Eur J Med Res: DNA Methylation Can Be Used as a Differential Tool for Liver Cancer Diagnosis and Preoperative Prediction of MVI Status

Recurrence is the main culprit affecting the long-term survival of hepatocellular carcinoma (HCC) patients, often occurring in the short term after surgery. Even small liver cancer (SHCC) has a 5-year recurrence rate of 50%. The important reason is that the tumor often invades the vasculature and easily forms sub foci in the liver. Postoperative specimen examination also shows that the infiltration of small liver cancer is not weak, and microvascular invasion (MVI) is often found under the microscope. Large liver cancer is often accompanied by visible tumor thrombi, with a MVI of 70%, indicating that MVI has appeared in the stage of small liver cancer.

Therefore, microvascular invasion (MVI) is one of the important risk factors for early postoperative recurrence of hepatocellular carcinoma. The presence of microvascular invasion (MVI)?suggests invasive behavior in hepatocellular carcinoma and is closely associated with increased risk of recurrence and decreased overall survival (OS).

At present, the diagnosis of MVI relies on postoperative histological examination, and there is still a lack of stable and effective predictive indicators for preoperative MVI status.

A recently published study suggests that?DNA methylation markers may be used to predict the MVI status of hepatocellular carcinoma before surgery.

This study included 35 liver cancer patients and 24 healthy subjects. Collect liver cancer tissue, matched normal liver tissue and plasma samples from 35 liver cancer patients, and collect plasma samples from 24 healthy subjects for whole genome methylation sequencing and methylation haplotype module (MHB) analysis.

Predictive models were constructed based on selected MHB markers with cross-validation.

Research results

1. DNA methylation markers can accurately distinguish between liver cancer tissue and normal liver tissue around the tumor

The DNA methylation levels of all MHBs were quantified using MHL, and unsupervised clustering analysis was performed based on MHL scores to visualize the degree of separation between liver cancer tissue and normal liver tissue surrounding the tumor. The results showed that based on the MHL score, liver cancer tissue was always separated from normal liver tissue around the tumor?(Figure 2A), indicating a significant difference in methylation levels between liver cancer tissue and normal liver tissue around the tumor.?Further principal component analysis was performed on MHL scores (Figure 2B), and consistent results were observed with unsupervised clustering analysis.

Using 65 selected MHBs and their MHL scores as independent variables, and liver tissue in the study queue as the training and validation sample set, two machine learning methods, Random Forest (RF) and Support Vector Machine (SVM), were used to train and cross validate the prediction model. The results showed that both the RF model and SVM model had high accuracy in distinguishing between hepatocellular carcinoma and normal liver tissue (RF model: AUC=0.98 (Figure 2D); SVM model: AUC=0.999).

2.?cfDNA methylation characteristics can accurately distinguish between hepatocellular carcinoma patients and healthy subjects

Researchers sequenced the methylation library of cfDNA samples from 35 liver cancer patients and 24 healthy subjects in the study queue to construct cfDNA methylation characteristics of hepatocellular carcinoma.

The results showed that the accuracy of distinguishing hepatocellular carcinoma patients and healthy subjects using cfDNA methylation features was high, with an AUC of 0.96.

3.?DNA methylation markers can predict MVI status

Train the RF MVI classification model on the training set samples using the selected 3 quantitative MHBs of MHL and 5 quantitative MHBs of UMHL as biomarkers to distinguish between tissue MVI (-) and MVI (+). In the cross validation set, the AUC of the model is 0.859; Indicating that DNA methylation markers can accurately distinguish between MVI (-) and MVI (+) in liver cancer tissue.

Based on the prediction of MVI status using DNA methylation markers, the recurrence free survival and overall survival of the MVI (-) group were significantly better than those of the MVI (+) group, which also confirms the high accuracy of DNA methylation markers in predicting MVI status.

Comparing the performance of DNA methylation markers and clinical features in predicting MVI status, it was found that DNA methylation markers had the highest accuracy in predicting MVI status, with an AUC of 0.915.?The accuracy of the AFP combined with DNA methylation markers model in predicting MVI status is not superior to that of individual DNA methylation markers (AUC=0.863).

Multivariate analysis showed that DNA methylation markers are independent risk factors for MVI.

Conclusion

This study found that methylation patterns can serve as a reliable diagnostic tool for liver cancer and preoperative prediction of MVI status.?The study screened and established cfDNA methylation markers for liver cancer detection and tissue DNA methylation markers for predicting MVI status, which may provide assistance for clinical doctors in formulating treatment strategies.

Reference:

1.?Liu Zhenyu, Wu Dan, Qu Jinling, Zeng Haifeng The Characteristics and Related Clinical Indicators of Microvascular Invasion in Liver Cancer [J] Journal of Hepatobiliary Pancreatic Surgery, 2017, 29 (2): 107-111

2. Hao Y, et al.?Identification of DNA Methylation Signatures for Hepatocellular Carcinoma Detection and Microvascular Invasion Prediction.?Eur J Med Res. 2022 Dec 5;27(1):276.