Flow diagram of the study.

Development and validation of tissue diagnosis model based on the 27 CRC-specific DMRs.

(A) The use of ROC curve analysis to assess the performance of the tissue diagnostic model in differentiating CRC patients from normal individuals of the training cohort (51 tissues we collected). (B-C) The use of ROC curve analysis to assess the performance of the tissue diagnostic model in differentiating CRC patients from normal individuals(B) and distinguish AA from Normal subjects (C) in tissue validation cohort 1(GSE 48684 dataset). (D-E) The sensitivity and specificity of the diagnosis model in the tissue training(D) and validation cohort 1(E).

Development and validation of a plasma diagnosis model based on the 27 CRC-specific DMRs.

(A-B) Heatmap illustrating the DMRs between CRC and advanced adenoma and healthy controls in the training (A) and validation cohort (B). (C) Methylation scores of Normal、AA and CRC individuals generated from the diagnosis model in the training and validation cohort (The dashed line represents the cutoff value). (D)The use of ROC curve analysis to assess the performance of the diagnosis model in differentiating CRC patients from normal individuals of the training cohort. (E) The use of ROC curve analysis to evaluate the performance of the diagnosis model in differentiating CRC patients from normal individuals of the validation cohort. (F-H) In the validation cohort, the performance of the diagnosis model in differentiating early-stage CRC (0-II) from normal individuals (F); in differentiating advanced-stage CRC (III-IV) from normal individuals (G); in differentiating AA from normal individuals (H). (I) The sensitivity and specificity of the diagnosis model stratified by stages in the training and validation cohort. Abbreviations: DMRs, differential methylated regions; ROC, receiver operating characteristic; AA, advanced adenoma; CRC, colorectal cancer.

External validation of plasma diagnosis model.

(A-C) In the external validation cohort, the performance of the diagnosis model in differentiating early-stage CRC (0-II) from normal individuals (A); in differentiating AA from normal individuals (B); in differentiating NAA from normal individuals (C). (D) Methylation scores of Normal、NAA、AA and CRC (0-II) individuals generated from the diagnosis model in the external validation cohort (The dashed line represents the cutoff value). (E) The sensitivity and specificity of the diagnostic model for NAA, AA, and CRC in the external validation cohort.

Development and validation of a plasma-based metastasis model and a plasma-based prognosis model.

(A-B) The methylation scores generated by the diagnosis model for CRC patients across various M stages (A) and different disease stages (B). (C) ROC curve analysis was performed to assess the performance of 27 DMRs methylation scores generated by the diagnosis model in distinguishing M1-CRC patients from M0-CRC patients in the training (AUC = 0.969) and validation (AUC = 0.955) cohorts. (D) ROC analysis evaluates the performance of 27 DMRs methylation scores generated by the diagnosis model in predicting the prognosis of CRC patients. (E) Kaplan-Meier survival curves comparing OS between the high-risk CRC group and low-risk CRC group of prognosis model in the training cohort and validation cohort. (F, H) Metastasis and prognosis prediction of the training and validation patitients (n = 88). The black dotted line at the cutoff value divides the patients into high-risk and low-risk groups. Yellow circles and gray circles represent patients without distant metastasis and with distant metastasis, respectively (F). Yellow circles and gray circles represent patients survived and deceased, respectively (H). (G, I) The proportion of metastasis and deceased is higher in the high-risk group. The p-value was calculated using a two-sided Fisher’s exact test. Abbreviations: DMRs, differential methylated regions; ROC, receiver operating characteristic; M1-CRC, colorectal cancer with distant metastasis; M0-CRC, colorectal cancer without distant metastasis; AUC, the area under the curve.

Methylation scores of paired tissue and plasma samples.

(A) The methylation scores of tissues from Normal, AA, CRC 0-II, CRC III-IV individuals. (B) The methylation scores of blood samples from Normal, AA, CRC 0-II, CRC III-IV individuals. (C) The mechanism diagram interpreting the methylation scores changes of tissue and plasma across different populations.

Screening for CRC-specific DMCs based on TCGA data.

(A) Venn diagram of the results from three DMCs filtering methods based on TCGA datasets. (B) Heatmap illustrating the methylation level of the 1438 methylated sites by unsupervised clustering. Abbreviations: DMCs, differential methylated CpG sites; CRC, colorectal cancer.

DMCs selection in tissues we collected.

(A) Volcano plot illustrates the CRC specific DMCs. (B) Heatmap analysis of the top 1400 DMCs between CRC tissues and adjacent tissues by unsupervised clustering. (C) Circular plot of 1400 DMCs depicting methylation level differences between CRC and adjacent tissues. (D) Distribution of 1400 DMCs in gene functional regions. Abbreviations: DMCs, differential methylated CpG sites; CRC, colorectal cancer.

Selection of methylation markers.

(A) The overlap of DMRs between CRC and adjacent tissues, as well as between CRC and normal tissues. (B) Heatmap illustrating the methylation level of the 404 DMRs in CRC and adjacent tissues. (C) The methylation level of the 404 DMRs in CRC and normal tissues. (D) The gene enrichment analysis for the 404 regions suggests their involvement in processes like cell-cell adhesion, digestive system development, and signaling pathways such as cAMP and cGMP. (E) Based on the results of plasma methylation detection, 142 DMRs were subsequently chosen for further analysis. (F) Heatmap illustrating the methylation levels of the 142 DMRs in the blood samples from normal individuals, those with AA, and patients with CRC. Abbreviations: DMRs, differential methylated regions; AA, adenomatous polyps; CRC, colorectal cancer.

The relationship between the methylation levels of the 27 DMRs and the transcription levels of their respective genes in TCGA datasets.

Development and validation of a plasma diagnosis model for discerning the specific staging of CRC patients.

(A) The use of ROC curve analysis to assess the performance of the diagnostic model in differentiating CRC 0-II patients from CRC III-IV patients of the training cohort. (B-C) The use of ROC curve analysis to evaluate the performance of the diagnostic model in distinguishing CRC 0-II patients from CRC III-IV patients (B), AA from CRC patients (C) of the validation cohort. (D-E) Kaplan-Meier survival curves comparing OS between the CRC III-IV group and CRC 0-II group in the training cohort (D) and validation cohort (E).

The construction of the metastasis model.

(A-C) The methylation scores generated from the diagnosis model for CRC patients with different age (A), gender (B) and location (C). (D) Methylation scores of M0-CRC and M1-CRC individuals generated from the metastasis model (The dashed line represents the cutoff value). (E-F) Kaplan-Meier survival curves comparing OS between the M1-CRC group and M0-CRC group of metastasis model in the training cohort (E) and validation cohort (F). (G) Functional enrichment analysis of genes associated with DMRs exhibiting methylation level changes in blood from patients with AA or CRC 0-II stages.

Explanation of specific phrase and abbreviation

Patient characteristics of the 15 paired tissues

Patient characteristics of the 64 participants (Tissues)

Patient characteristics of the blood cohort

Annotation information for 27 regions

Univariate and multivariate analysis of methylation scores and clinical parameters.

Comparison of diagnostic performance between the methylation score and CEA for early-stage colorectal cancer and adenomas in the external validation cohort.