Skip to main page content
Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2018 Jan 18;10:8.
doi: 10.1186/s13148-017-0436-1. eCollection 2018.

The Signature of Liver Cancer in Immune Cells DNA Methylation

Affiliations
Free PMC article

The Signature of Liver Cancer in Immune Cells DNA Methylation

Yonghong Zhang et al. Clin Epigenetics. .
Free PMC article

Abstract

Background: The idea that changes to the host immune system are critical for cancer progression was proposed a century ago and recently regained experimental support.

Results: Herein, the hypothesis that hepatocellular carcinoma (HCC) leaves a molecular signature in the host peripheral immune system was tested by profiling DNA methylation in peripheral blood mononuclear cells (PBMC) and T cells from a discovery cohort (n = 69) of healthy controls, chronic hepatitis, and HCC using Illumina 450K platform and was validated in two validation sets (n = 80 and n = 48) using pyrosequencing.

Conclusions: The study reveals a broad signature of hepatocellular carcinoma in PBMC and T cells DNA methylation which discriminates early HCC stage from chronic hepatitis B and C and healthy controls, intensifies with progression of HCC, and is highly enriched in immune function-related genes such as PD-1, a current cancer immunotherapy target. These data also support the feasibility of using these profiles for early detection of HCC.

Keywords: DNA methylation; Hepatocellular carcinoma; Immune functions; Peripheral white blood cells.

Conflict of interest statement

Informed consent has been obtained from all participants and the study received ethical approval from The Capital Medical School in Beijing and McGill University (IRB Study Number A02-M34-13B).Participants/patients have given their consent for their data to be published in the report.The authors declare that MS, YZ, SP, and NL have applied for patent protection, MS has equity in HKG epitherapeutics and Montreal epiterapia, and DC has equity in Montreal epiterapia.Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Figures

Fig. 1
Fig. 1
Correlation between quantitative distribution of site-specific DNA methylation levels and progression of HCC. a A genome wide view (IGV genome browser) of the escalating differences in DNA methylation from healthy controls (delta beta) in 3924 CG sites whose quantitative levels of methylation correlate with HCC progression (r > 0.8, r < − 0.8; delta beta > 0.2, < − 0.2; p < 10−7) in PBMC from HCC and hepatitis B and C patients. HepB-Hepatitis B; HepC-Hepatitis C; CAN1-stage 1 HCC; CAN2- Stage 2 HCC: CAN3- Stage 3 HCC; CAN4-Stage 4 HCC. b Box plot of DNA methylation delta beta values of the 3924 CG sites whose levels of methylation correlate significantly (p < 10−7) with HCC progression. Sites that are hypomethylated relative to healthy control during progression of HCC (upper panel) and sites that are hypermethylated relative to healthy controls (bottom panel) are shown separately. c Heat map of hierarchical clustering using one minus Pearson correlation of 69 people by DNA methylation beta values of the 3924 CG sites
Fig. 2
Fig. 2
Differentially methylated CG sites at different stages of HCC and “cross-validation.” a Number of CG sites that are differentially methylated between different stages of HCC and healthy controls (p < 10−7) green: hypomethylated, red: hypermethylated. b Heat map presentation of hierarchical clustering of 69 people by 74 differentially methylated CGs between HCC stage 1 and control. c Heat map of hierarchical clustering of 69 people by 298 differentially methylated CGs between HCC stage 4 and control. d Ven diagram of the overlap between differentially methylated CG sites at different HCC stages (1–4). Significance was determined using Fisher exact test for all overlaps and all overlaps were highly significant (p < 1.7 × 10−321)
Fig. 3
Fig. 3
Staging of HCC using differentially methylated CGs. a Heat map presentation of hierarchical clustering of 69 people by 350 non-redundant CGs that are differentially methylated between different HCC stages and healthy controls. b Prediction of late stage HCC in 69 patients using a penalized model trained on a randomized half of the HCC patients and controls (“training set”) and tested on the other half (“validation set”). The plot shows all samples (the training and validation sets combined). The y axis indicates the predicted probability of late stage HCC for each person (from 0 to 1). c Prediction of late stages of HCC using the penalized model in T cell samples
Fig. 4
Fig. 4
Differences in DNA methylation between HCC and healthy controls in T cells DNA overlap with differences in methylation in PBMC. a Heat map presentation of hierarchical clustering of ten healthy controls and HCC samples from T cells by 370 significantly differentially methylated CGs. b Heat map presentation of hierarchical clustering of PBMC DNA methylation samples from 69 people by 370 CGs “trained” in T cells. c Heat map presentation of hierarchical clustering of T cell samples from ten healthy controls and ten HCC by 350 CGs “trained” on T cell DNA. d Overlap of differentially methylated CGs in HCC in T cells and differentially methylated CGs in PBMC at different stages of HCC
Fig. 5
Fig. 5
Progressive hypomethylation of PD-1 gene during HCC progression. a Correlation (linear fit) between average beta values for cg14453145 positioned at the TSS region of PD-1 and control (CTRL) (stage code 0), hepatitis B (HepB) (1), hepatitis C (HepC) (2), and the four stages of HCC (St_1 to St_4) (3–6) diagnoses (equation and R values are indicated). One way ANOVA showed a highly significant effect of diagnosis on DNA methylation (p = 1 × 10−13; F = 20.77). Bonferroni adjusted pairwise comparison revealed significant differences between HCC stage 1 and control (p = 0.0058) and hepatitis B (p = 0.00079); between stage 2 and control (p = 0.0004) and hepatitis B (p = 4.9 × 10–5); between stage 3 and control (p = 4.8 × 10–9), hepatitis B (p = 4.9 × 10–10), hepatitis C (p = 1.8 × 10–5) and stage 1(p = 0.00993); between stage 4 and control (p = 2.1 × 10–8), hepatitis B (p = 2.3 × 10−9), hepatitis C (p = 5.9 × 10−5) and stage1 (p = 0.00558). b Correlation (linear fit) between average beta values of 24 CGs associated to the PD-1 gene on the Illumina 450k arrays that were hypomethylated in HCC (average methylation score was calculated per person, the average of these scores were then calculated per group). There was a highly significant effect of diagnosis on DNA methylation as determined by one way ANOVA (p = 2.2 × 10–16, F = 52.74). Bonferroni adjusted pairwise comparison revealed significant differences between stage 1 HCC and control (p = 1.2 × 10−7), hepatitis B (p = 7 × 10−8), hepatitis C (p = 0.00487), stage 3 (p = 0.00081), and Stage 4 (p = 6.4 × 10−6); between stage 2 and control (p = 4.7 × 10−11), hepatitis B (p = 2.8 × 10−11), hepatitis C (p = 3.8 × 10−6), and stage 4 (p = 0.00645); between stage3 and control (p = 3.4 × 10−15), hepatitis B (2.1 × 10−15), hepatitis C (2.1 × 10−10), and stage 1 (p = 0.00081); and between stage 4 and control (2 × 10−16), hepatitis B (2 × 10−16), hepatitis C (1.8 × 10−12), stage 1 (6.4 × 10−6), and Stage 2 (p = 0.00645). c Heat map presentation of hierarchical clustering (city block) of 69 people by 5’DMR whose average methylation correlates with HCC progression
Fig. 6
Fig. 6
Validation of differentially methylated CGs in the discovery set and validation set by pyrosequencing. a Top row, CG sites that are differentially methylated between HCC (n = 10) and healthy controls (n = 10) in T cells (significance was measured by student t test set at a threshold of < 0.05). The primers for pyrosequencing and conditions are listed in Additional file 18: Table S17. The scattered plot shows the mean and 95% confidence intervals (C.I.). The average methylation for three CG sites in the SLFN14 differentially methylated region is shown in the left panel. Summary of statistics including CI, SD, and SEM values are presented in Additional file 16: Table S15. b Validation by pyrosequencing of DNA extracted from T cells in the validation set. ANOVA was used to compare variance between the hepatitis B (HepB) control and other groups healthy (n = 10), hepatitis B (n = 10), hepatitis C (HepC) (n = 10) group and the HCC stages 1 (n = 8), 2 (n = 12), 3 (n = 8), and 4 (n = 22). STAP1 replication presents pyrosequencing data from T cells DNA from the second replication cohort (Additional file 15: Table S14). c ROC curve measuring specificity (Y axis) and sensitivity (X axis) of STAP1 methylation as a biomarker for discriminating HCC from healthy controls in T cells first cohort (Illumina 450 K data), in first validation set (pyrosequencing) and third validation set (pyrosequencing replication). d. ROC curve for STAP1 methylation as a biomarker for distinguishing HCC from healthy persons and chronic hepatitis in PBMC (Illumina), first validation set (pyrosequencing), and third validation set (pyrosequencing, replication). Statistic code: * 0.05, ** 0.01, *** 0.001

Similar articles

See all similar articles

Cited by 9 articles

See all "Cited by" articles

References

    1. Blair GE, Cook GP. Cancer and the immune system: an overview. Oncogene. 2008;27:5868. doi: 10.1038/onc.2008.277. - DOI - PubMed
    1. Ehrlich P. Ueber den jetzigen Stand der Karzinomforschung. Ned Tijdschr Geneeskd. 1909;5:273–290.
    1. Vesely MD, Kershaw MH, Schreiber RD, Smyth MJ. Natural innate and adaptive immunity to cancer. Annu Rev Immunol. 2011;29:235–271. doi: 10.1146/annurev-immunol-031210-101324. - DOI - PubMed
    1. Dunn GP, Bruce AT, Ikeda H, Old LJ, Schreiber RD. Cancer immunoediting: from immunosurveillance to tumor escape. Nat Immunol. 2002;3:991–998. doi: 10.1038/ni1102-991. - DOI - PubMed
    1. Swann JB, Smyth MJ. Immune surveillance of tumors. J Clin Invest. 2007;117:1137–1146. doi: 10.1172/JCI31405. - DOI - PMC - PubMed

Publication types

MeSH terms

Substances

Feedback