Detection and clinical significance of CEACAM5 methylation in colorectal cancer patients

Abstract Colorectal cancer (CRC) is a globally common cancer, and the serum carcinoembryonic antigen (sCEA) is widely applied as a diagnostic and prognostic tumor marker in CRC. This study aimed to elucidate the mechanism of CEA expression and corresponding clinical features to improve prognostic assessments. In CRC cells, hypomethylation of the CEACAM5 promoter enhanced CEA expression in HCT116 and HT29 cells with 5‐aza‐2′‐deoxycytidine (5‐Aza‐dC) treatment. Our clinical data indicated that 64.7% (101/156) of CRC patients had an sCEA level above the normal range, and 76.2% (77/101) of those patients showed a lower average CpG methylation level of the CEACAM5 promoter. The methylation analysis showed that both CRC cell lines and patient samples shared the same critical methylation CpG regions at −200 to −500 and −1000 to −1400 bp of the CEACAM5 promoter. Patients with hypermethylation of the CEACAM5 promoter showed features of a BRAF mutation, TGFB2 mutation, microsatellite instability‐high, and preference for right‐sided colorectal cancer and peritoneal seeding presentation that had a similar clinical character to the consensus molecular subtype 1 (CMS1) of colorectal cancer. Additionally, hypermethylation of the CEACAM5 promoter combined with evaluated sCEA demonstrated the worst survival among the patients. Therefore, the methylation status of the CEACAM5 promoter also served as an effective biomarker for assessing disease prognosis. Results indicated that DNA methylation is a major regulatory mechanism for CEA expression in colorectal cancer. Moreover, our data also highlighted that patients in a subgroup who escaped from inactivation by DNA methylation had distinct clinical and pathological features and the worst survival.


| INTRODUC TI ON
Colorectal cancer is the third most common cancer and the fourth leading cause of cancer deaths in the world. 1 In Taiwan, CRC was the second most common cancer type and the third leading cause of cancer deaths in 2017. 2 CRC has unique features such as several known genetic variations, genomic instability, and a CIMP. 3,4CRC tumorigenesis is highly related to those genetic and epigenetic variations.Chromosomal instability and MSI are being researched to clarify the pathogenesis and prognosis of CRC.Aberrant DNA methylation, through the CIMP, enhances DNA hypermethylation at promoter CpG islands of tumor suppressor genes or other tumor-related genes, leading to transcription inactivation and gene silencing. 5erefore, the CIMP is considered an early event characteristic of the serrated pathway of colorectal tumorigenesis. 6 clinical practice, surgery is the primary treatment modality for CRC followed by radiation and chemotherapy, but advanced CRC still has poor survival outcomes.Therefore, in order to detect disease recurrence at an early stage, CRC surveillance is of utmost importance in clinical practice.This includes regular monitoring of images, such as imaging scans, as well as the utilization of valuable biomarkers.Early diagnosis through effective surveillance can significantly improve patient outcomes by enabling timely intervention and management of recurrent CRC.sCEA is a well known tumor marker for CRC, initially found on cell membranes of gastrointestinal tract cells during the embryonic period but decreasing after maturation. 7CEA has an immunoglobulin-like structure and many glycosylation modification sites belonging to a group of CEA-related CAMs (CEACAMs) containing 12 proteins (CEACAM1, -3 to -8, -16, -18 to -21). 8][10] In CRC, Gold and Freedman discovered CEA expression in colon cancer tissues that serves as a tumor marker in CRC. 11A clear correlation of tumor metastasis with sCEA in CRC was proven through in vivo and in vitro studies. 12,13sCEA expression was correlated with the CRC prognosis and was mainly used for disease follow-up and treatment response indicators. 14Serial measurements of sCEA are widely recommended in surveillance; however, agreement is lacking about what constitutes clinically significant changes in sCEA levels. 15Many clinical features have shown that the sCEA could predict the disease prognosis, severity of the disease, and response to therapy. 16,17CRC patients with elevated sCEA tend to have the potential for liver metastasis, and the probability is highly correlated with the sCEA level.9][20] Moreover, the CEA expression regulatory mechanism during CRC progression is still unclear.Therefore, elucidating the CEA regulatory mechanism may improve the application of sCEA in clinical diagnoses.
Epigenetic regulation controls gene expressions through DNA methylation, histone modifications, and chromatin remodeling.
Abnormal methylation changes of CpG islands of a tumor suppressor may be used as one of the available means for the early detection of tumor patients. 21Tran et al. first indicated the correlation of DNA hypomethylation with CEA expression in CRC cell lines. 22wever, the sCEA level, DNA methylation pattern, and CRC's clinical characteristics were not addressed.To extend the understanding and improve sCEA practice in clinical applications, we conducted next-generation sequencing (NGS)-based methylation sequencing to profile the DNA methylation pattern of CEACAM5 and analyze its correlations with clinical features in CRC samples.

| Cell lines
Two human CRC cell lines, HCT116 and HT29, were used in this study.The cell lines were obtained from the American Type Culture Collection (Manassas, VA, USA) and were authenticated before experiments were performed.These cells were cultured in Dulbecco's modified Eagle's medium (DMEM; Gibco/Life Technologies) supplemented with 10% fetal bovine serum (FBS, Gibco/Life Technologies) and 100 U/mL of penicillin.Cells were cultured at 37°C in a 5% CO 2 incubator.

| Clinical CRC samples
Surgically resected colon tissues with neoplastic and non-neoplastic areas were obtained after CRC surgery.A sample was collected after a patient signed an informed consent form, and the protocol was approved by the institutional review board (IRB) of Taipei Veterans General Hospital (TVGH; IRB no.2019-01-016BC).
According to the pathology reports, the clinicopathological indicated that DNA methylation is a major regulatory mechanism for CEA expression in colorectal cancer.Moreover, our data also highlighted that patients in a subgroup who escaped from inactivation by DNA methylation had distinct clinical and pathological features and the worst survival.

| 5-Aza-2′-deoxycytidine (5-Aza-dC) treatment of cell lines
The HCT116 and HT29 colon carcinoma cell lines were seeded the day before treatment and then treated with 5-aza-dC for 48 h.The treated concentration of 5-aza-dC was 10 μM for HCT116 and 15 μM for HT29 cells.After treatment, cells were harvested and washed with 1× phosphate-buffered saline (PBS) before genomic RNA and DNA extraction for the CEA expression and DNA methylation surveys.

| Extraction of DNA and quantification of CEA messenger (m)RNA
DNA from cell lines and colon tissues was extracted using a DNA extraction kit system (PicoPure DNA extraction kit, MDS Analytical Technologies) for the DNA methylation survey.Total RNA was extracted from cell lines using an RNeasy Mini kit (Qiagen).Total RNA (1 μg) was reverse-transcribed with a high-capacity RNA-tocomplementary (c)DNA Kit (ThermoFisher Scientific).A real-time reverse-transcription polymerase chain reaction (PCR) was performed on a Roche LightCycler® 96 System using FastStart Essential DNA Green Master with the primers forward: CTGGC CGC AAT AAT TCC ATAG and reverse: CCAGC TGA GAG ACC AGG AGAA for CEA mRNA measurements.

| Quantitative DNA methylation analyses by next-generation sequencing
The promoter methylation status of CEACAM5 was determined by NGS on bisulfite-treated genomic DNA.Extracted DNA (100 ng) from the cell lines and clinical tissues was used for bisulfite treatment and methylation evaluation.A Zymo EZ DNA methylation Lightning Kit (Zymo Research) was used for bisulfite conversion, followed by an EpiGnome™ Kit (Epicenter) to prepare the bisulfite sequencing libraries before sequencing.Sequencing was performed with an Illumina HiSeq2500 genome sequencer (USA) to determine the reads of thymine and cytosine in each CpG site in the sequence.

| Statistical methods
Statistical analyses were performed using SPSS version 24.0 software (SPSS, Chicago, IL, USA).Data are shown as the prevalence, mean (standard deviation (SD)), or median (range).Discrete variables were compared using a chi-squared test or Fischer's exact test as appropriate.Continuous variables were compared using the Mann-Whitney U-test.The overall survival (OS) was displayed with Kaplan-Meier survival curves and compared using the log-rank test.
All statistical tests were two-sided, with the threshold for significance set to p < 0.05.We next compared the genetic variants and clinical features of these primary tumors.Results indicated that 26 CEA is a cell surface glycoprotein used as a clinical biomarker for gastrointestinal cancers, especially colorectal malignancies.It promotes tumor development through its role as a cell adhesion molecule. 23An sCEA level of >5.0 ng/mL is considered positive. 24 these samples, 55 patients (35.3%) were categorized as normal (<5 ng/mL), and 101 patients (64.7%) had elevated sCEA levels of >5 ng/mL.The overall median sCEA level was 7.83 ng/mL.Since DNA methylation regulation is a critical epigenetic regulation in CRC progression and shows an inverse correlation with CEA TA B L E 1 Clinical characteristics of colorectal cancer patients.

| Survey of CEA levels and DNA methylation statuses in CRC cell lines
We first examined CEA expression in two CRC cell lines.
Quantitative PCR (qPCR) results showed that the CEA messenger (m)RNA level was significantly higher in HT29 compared to HCT116 cells (p < 0.001; Figure 1A).Next, we analyzed the DNA methylation pattern of CpG sites in the CEACAM5 promoter by NGS (Figure 1B).Next, we examined whether the DNA methylation status influences CEA expression.We treated CRC cells with 5-aza-2′-deoxycytidine (5-Aza-dC).5-Aza-dC is incorporated into nucleic acids and prevents methylation at CpG sites by irreversible covalent binding to DNMT1, leading to loss of methyltransferase activity and demethylation of DNA. 25 Results indicated that, on average, methylated CEA promoters decreased from 27.6% to 21.3%, and the mRNA of CEACAM5 increased in HT29 cells after 5-Aza-dC treatment (Figure 2A).Consistently, HCT116 cells treated with 5-Aza-dC showed that the average methylated promoter decreased from 38.6% to 25.2%, and CEA mRNA also increased (Figure 2B).Moreover, the NGS-based methylation sequencing data indicated that the decrement in the methylated region in both

| Investigation of the CEACAM5 methylation pattern in CRC tumors
We next examined the methylation status of the CEACAM5 promoter in 156 CRC clinical samples and the corresponding adjacent normal tissues.Results indicated that 73.7% (115/156) of CRC tumors had decreased average CpG methylation over the CEACAM5 promoter compared with the adjacent normal part, and 26.3% (41/156) of cases belonged to the increased CEACAM5 methylation group (Table 2).Among the sCEA-increased patients, 76.2% (77/101) of patients had lower CpG methylation of the CEACAM5 promoter, and 23.8% (24/101) of patients had higher CpG methylation of the CEACAM5 promoter (Tables 3 and 4).Although the tumor CEACAM5 promoter methylation percentage in the elevated sCEA level group did not show significant difference compared with the normal sCEA group (p = 0.478; Figure 3A), when excluding 24 patients who escaped from inactivation by DNA methylation (those with increased CEACAM5 promoter methylation and elevated sCEA), the increased sCEA expression group showed a significantly lower CEACAM5 promoter methylation percentage (p = 0.016; Figure 3B).The relationship between tumor CEACAM5 promoter methylation and the corresponding sCEA can be seen in Figure S4.There was a borderline significant negative correlation between sCEA and CEACAM5 promoter methylation (p = 0.054).
The results indicated that, in most CRC patients, increased sCEA expression is associated with a decrease in the methylation level of the CEACAM5 promoter.This result was consistent with our finding of the DNA methylation hot spots in the CRC cell lines (Figures S2 and S3).Next, we compared the methylation ratio between the paired tumor and normal samples.Results showed that the average CEACAM5 promoter methylation percentage in the tumor part was 17% ± 6%, whereas that of a matched adjacent normal part was 20% ± 3% (p < 0.001; Figure 4C).However, not all tumor parts had decreased CEACAM5 promoter methylation levels than the paired normal samples.
Interestingly, a consistent methylation pattern was noted in normal  S5).

| The clinical significance of sCEA level and CEACAM5 promoter methylation in CRC patients
We analyzed the clinical features of patients who had increased or decreased CEACAM5 promoter methylation and sCEA levels.
Intriguingly, 26.3% (41/156) of patients with higher methylated CEACAM5 promoter levels possessed clinical features such as a BRAF mutation, TGFB2 mutation, MSI-H, RCC, or recurrent peritoneal seeding compared to patients with lower CEACAM5 methylated promoter levels (Table 2).In CRC cases with higher sCEA levels, clinical features showed a higher proportion of loss of the 18q mutation, more stage IV and synchronous liver metastasis, and poorer OS (Table 3).Among these patients, 23.8% (24/101) of them had higher methylation of the CEACAM5 promoter accompanied by increased CEA expression, indicating that the CEA expression is not completely regulated by DNA methylation (Table 4).These patients exhibit features such as BRAF mutations, TGFβ2 mutations, MSI-H, right-sided CRC, poor tumor differentiation, and recurrent peritoneal seeding.Notably, these features were not observed in the CEA elevation group.Those CRC patients also showed similar clinical features as the higher CEACAM5 methylation group (Tables 2   and 4).Altogether the hypermethylated CEACAM5 group showed these clinical features resemble the CMS1 classification of CRC with a BRAF mutation, TGFB2 mutation, MSI-H, and proximally located colon tumor.promoter and normal sCEA showed the best OS.There was no significant difference in the OS between patients with hypomethylated CEACAM5 promoter and elevated sCEA and those with hypermethylated CEACAM5 promoter and normal sCEA (Figure 5).
The multivariable analysis, which considered factors such as age, gender, tumor location, BRAF mutation, MSI status, KRAS mutation, TGFβ2 mutation, and poor tumor differentiation, indicated that the combination of CEACAM5 methylation and sCEA levels independently affected disease prognosis (Table 5, p = 0.023) The results indicated that integration of sCEA and CEACAM5 promotor methylation presents a significant and informative approach for assessing disease prognosis.

| DISCUSS ION
CEA is a glycoprotein found by Gold and Freedman in colon cancer tissues, which was then applied as a CRC tumor marker. 26CEA is a cell membrane protein and can be cleaved by phospholipase C and phospholipase D and released into the circulation. 27sCEA expression was correlated with the CRC prognosis and was mainly used for disease follow-up and as a treatment response indicator.Our previous study also indicated that 25.6% of stage IV CRC patients did not have elevated sCEA. 18Consistently, previous surveys showed that around 30% of metastatic CRC cases had no elevated sCEA. 19,20A recent study indicated that tissue (t)CEA expression rather than sCEA is an independent factor associated with a poorer CRC prognosis in stages I-III of CRC. 28However, the post-translational process of releasing sCEA and the effect of DNA methylation regulation on tCEA remain unresolved.
In this study, we evaluated the influence of DNA methylation on regulating sCEA expression.Our data indicated that the sCEA (A) The CEACAM5 promoter methylation percentages of the sCEA at >5 ng/mL and <5 ng/mL.(B) CEACAM5 promoter methylation percentages between the sCEA at >5 ng/mL and <5 ng/mL with the exclusion of 24 patients whose sCEA had escaped inactivation by DNA methylation.See also Tables 3 and 4.
activation through hypermethylation-induced transcriptional activation. 30These data indicated that epigenetic contributions to transcriptional regulation occur in a more complex and more dy- CRC patients could benefit from ICIs. 31,32Whether tumors with hypermethylated CEACAM5 share the same clinical characteristics as CMS1 tumors warrants further investigation.In clinical applications, previous studies have examined genetic methylation in CRC, identifying several genes with methylation patterns that serve as markers for CRC. 33Our data demonstrates that combining the assessment of CEACAM5 promoter methylation status with sCEA levels provides a more comprehensive understanding of disease prognosis, highlighting the potential utility of CEACAM5 promoter methylation level as a marker in clinical settings.Our study reveals that CRC patients with increased CEACAM5 promoter methylation and elevated sCEA levels exhibit the poorest prognosis (Figure 5).
regulation, CEACAM5, colorectal cancer, DNA methylation features of these samples and patients were collected in TVGH from August 2010 to May 2016.All subjects had been diagnosed with CRC.The diagnostic pathology reports were performed using paraffin-embedded sections combined with an immunohistochemical (IHC) study of BRAF and MMR proteins to know the BRAF and MSI status.Tissues were submerged in an RNAlater solution (ThermoFisher Scientific) for further DNA extraction.Clinical CEA measurements were performed at the Department of Pathology and Laboratory Medicine of TVGH using an electrochemiluminescence immunoassay (ECLIA) method with an analytical sensitivity of 0.3 ng/mL.
The proportion of reads between thymine and cytosine in the CpG sites can show the percentage of methylation of CpG sites.The methylation of CpG sites in the region from 1 to −1800 bp relative to the transcription start site of the CEACAM5 promoter sequence was analyzed.HCT116 DKO non-methylated DNA and human HCT116 DKO methylated DNA in the Human Methylated and Non-methylated DNA Set (Zymo Research) were respectively used as the positive and negative controls in the methylation survey.The methylation level was calculated from reads between the selected CEACAM5 promoter sequence on each CpG site with reads of thymine and cytosine.
Mutation analyses of KRAS codons 12 and 13 and BRAF codon 600 were performed by pyrosequencing at the pathology department for clinical requirements.Genomic DNA was amplified by a PCR and sequenced with the PyroMark™ KRAS kit and the PyroMark™ BRAF kit according to the manufacturer's instructions.Results were part of cancer pathology reports and contained relevant information for clinical management.Other colon cancer genetic mutation evaluations were performed by MassArray with hotspots reported by the COSMIC database, and 139 mutations in 12 genes were checked.The PCR for mutation detection was designed by MassArray Assay Design 3.1 software (Sequenom), and DNA products were analyzed by the MassArray Analyzer 4 system (Sequenom) and Typer 4.0 software (Sequenom) to detect mutations.
This study included 156 Taiwanese CRC samples from the TPE-VGH Biobank from August 2010 to May 2016.The clinical characteristics of the patient cohort are shown in Table 1.There were 64.7% (101/156) male patients and 35.3% (55/156) female patients, and the median age at diagnosis was 69 (range 35-92) years.Additionally, 26.3% (41/156) of the patients had a smoking history, while 73.7% (115/156) did not.The observed increase in sCEA levels shows no significant difference between the smoking group and the nonsmoking group (61% vs. 50.4%,p = 0.227).Among them, 23.1% (36/156) of patients were diagnosed with stage IV CRC and had received palliative primary tumor resection, among which 61.1% (22/36) of patients had liver metastasis, 27.8% (10/36) of patients had lung metastasis, and 19.4% (7/36) of patients had peritoneal seeding disease among these 36 stage IV CRC patients.The rest of the patients in stages I-III received surgical resection of the tumor lesion under curative intent.Figure S1 illustrates the adjuvant therapy and recurrence patterns in patients with stages I-III CRC.For stage I CRC, one patient underwent adjuvant chemotherapy due to an unclear resection margin.Of these patients, one out of 14 experienced lung metastasis during the follow-up period.For stage II CRC, 10 patients received adjuvant chemotherapy based on high-risk factors as per the National Comprehensive Cancer Network (NCCN) guidelines.Within this group, there were two instances of local recurrence, three of lung metastasis, and one of peritoneal seeding during the follow-up.In stage III CRC, 79.3% of patients received adjuvant chemotherapy according to the NCCN guidelines.The remaining patients declined adjuvant chemotherapy due to personal reasons or poor health conditions.Among these stage III patients, nine developed liver metastasis, seven had peritoneal seeding, six had lung metastasis, three experienced local recurrence, and three had bone metastasis.The overall recurrence rate for stage III CRC in this study was 34.5%.

|
275   expression,22 this implies that DNA methylation may control CEA expression.Although studies indicated that CEA expression was correlated with DNA hypomethylation in colorectal cell lines, the DNA methylation status of CEACAM5 has not previously been evaluated in clinical samples.Therefore, we evaluated the CEACAM5 promoter methylation status and sCEA expression among the different stages of CRC patients and CRC cell lines to improve our understanding of sCEA's role in diagnoses.

F I G U R E 2
Effect of 5-Aza-dC on DNA methylation of the CEACAM5 promoter and corresponding CEACAM5 expression of colorectal cancer (CRC) cell lines.(A) Left: CEACAM5 promoter methylation percentages before and after 5-Aza-dC treatment in HT29 cells.Right: Corresponding CEACAM5 mRNA expressions in control and 5-Aza-dCtreated cells.Data are presented as the mean ± SD. n = 3 independent experiments (each experiment contained two technical replicates).(B) Left: CEACAM5 promoter methylation percentages before and after 5-Aza-dC treatment in HCT116 cells.Right: Corresponding CEACAM5 mRNA expressions in control and 5-Aza-dCtreated cells.Data are presented as the mean ± SD. n = 3 independent experiments (each experiment contained two technical replicates).*p < 0.05.See also Figures S2 and S3.F I G U R E 1 DNA methylation patterns of CpG sites in the CEACAM5 promoter of colorectal cancer (CRC) cell lines.(A) RT-qPCR for analyzing the relative expression of CEACAM5 in HCT116 and HT29 cells.Data are presented as the mean ± SD. n = 3 independent experiments (each experiment contained two technical replicates).(B) NGS analysis results of CpG methylation distribution in the CEACAM5 promoter of HCT116 and HT29 cells.(C) Average percentages of CpG methylation of the CEACAM5 promoter in HCT116 and HT29 cells.(D) Average percentages of CpG methylation of HCT116 and HT29 cells in the promoter at −200 to −500 and −1000 to −1400 bp from the transition start site (TSS) of CEACAM5.**p < 0.01.Results showed that methylation of the CEACAM5 promoter of HT29 cells (27.6%) was lower than that of HCT116 cells (38.6%), which indicated that hypomethylation of the CEACAM5 promoter may enhance CEA mRNA production in CRC cell lines (Figure 1C).Moreover, sequencing data indicated that two central regions of CpG sites, of −200 to −500 (two CpG sites) and −1000 to −1400 bp (eight CpG sites) in the CEACAM5 promoter region, showed the greatest difference between the two cell lines (Figure 1B,D).

5 -
Aza-dC-treated CRC cell lines occurred at −200 to −500 and −1000 to −1400 bp of the CEACAM5 promoter region (Figures S2 and S3).These results indicated that the regions of −200 to −500 and −1000 to −1400 bp on the CEACAM5 promoter are hot spots for DNA methylation changes.

Furthermore, DNA methylation
profiling based on the NGS analysis of 156 patients indicated that the methylation difference in matched normal-tumor pairs was located in regions −200 to −500 and −1000 to −1400 bp on the CEACAM5 promoter (Figure 4A,B).
level was mainly regulated by DNA methylation control of the CEACAM5 promoter.Traditionally, epigenetic reprogramming in cancer contributes to cancer development by directly inhibiting gene expressions through promoter hypermethylation or modification, particularly TFs. 29Therefore, mining associated TFs that can bind to these methylation hot spots and examining whether critical TFs are lost may explain why those patients had hypomethylation but low sCEA expression.Individual differences in those critical TFs may cooperate with DNA hypomethylation and regulate CEA expression within tumor samples.Interestingly, a subgroup of CRC patients had increased sCEA expression and had escaped from canonical regulation of gene inactivation by DNA methylation.Emerging studies showed that promoter hypermethylation is associated with gene F I G U R E 3 Serum carcinoembryonic antigen (sCEA) levels and corresponding DNA methylation percentages of the CEACAM5 promoter.
namic manner.However, the molecular mechanism of hypermethylation-induced gene activation is currently unclear.Specific TFs and hypermethylation enhance gene activation under specific contexts showing that a more detailed investigation of the complex epigenetic regulation is warranted.Moreover, our data indicated that the hypermethylated CEACAM5 group showed molecular pathological features with a F I G U R E 4 DNA methylation pattern of CpG sites in the CEACAM5 promoter of colorectal cancer (CRC) patients.(A) NGS analytical results of the CpG methylation distribution of 156 CRC patients.(B) Average percentages of the CpG methylation pattern of the promoter at −200 to −500 and −1000 to −1400 bp from the TSS of CEACAM5.(C) Average CpG methylation percentages in 156 CRC patients.See also Figure S5.BRAF mutation, TGFB2 mutation, MSI-H, and proximally located colon tumors that were similar to CMS1 tumors.CMS1 tumors are enriched with activated Th1 lymphocytes, cytotoxic T cells, natural killer (NK) cell infiltration, and upregulated immune checkpoints such as programmed death ligand (PD)-1.Therefore, CMS1

However, the precise
molecular mechanism through which hypermethylation induces gene activation, including the specific activation of the CEACAM5 gene and its clinical correlation with CRC prognosis, remains unclear.It is hypothesized that hypermethylation may impede the binding of repressive TFs and distal regulatory elements.30,34Interestingly, hypermethylation-induced gene activation has been observed in various contexts, such as induced pluripotent stem cells (iPSCs), early development, and malignancy.This raises the question of whether DNA methylation, as a potential novel pathway, instigates gene expression changes that drive malignancies to adopt a more pluripotent phenotype.In our study, this CRC subgroup displaying this trend toward a higher frequency of BRAF mutations (21.7%) and TGFB2 mutations (12.5%) consisted of predominantly right-sided CRC cases (50%) and exhibited a higher risk of recurrent peritoneal seeding (12.5%).All these factors have been associated with a poor prognosis for CRC.Based on our data, it can be inferred that hypermethylation accompanied by CEA activation signifies the worst clinical outcome among these patients.Those with a poor prognosis require more aggressive monitoring for recurrent disease in stages I-III for CRC, especially for peritoneal seeding, and a more intensive treatment approach in stage IV disease.In conclusion, DNA methylation is the major regulatory mechanism governing sCEA expression in CRC, and hypomethylation could enhance sCEA expression.Furthermore, our data also identified two central regions of CpG sites at −200 to −500 and −1000 to −1400 bp in the CEACAM5 promoter region, which are vital for regulating sCEA expression.Moreover, a subgroup of patients with hypermethylated CEACAM5 promoters that escape from inactivation by DNA methylation demonstrated the molecular and clinical features with a BRAF mutation, TGFB2 mutation, MSI-H, recurrent peritoneal seeding, and worst prognosis, which may provide new insights into CRC.F I G U R E 5 Kaplan-Meier curves of overall survival.The overall survival analysis in CRC patients according to CEACAM5 promoter methylation status and sCEA level.Taipei Veterans General Hospital (TPEVGH IRB no.2019-01-016 BC) and was performed in accordance with the Declaration of Helsinki.Informed Consent: N/A.Registry and the Registration No. of the study/trial: N/A.

Patient characteristics (N = 156)
FIGURE 1 Legend on next page Clinical characteristics between CEACAM5 promotor methylation status of CRC patients with CEA elevated.Clinical characteristics between difference CEA level of CRC patients.We next investigated the prognostic impact of sCEA combined with CEACAM5 promoter methylation in CRC patients.The level of TA B L E 4 TA B L E 3