Karolinska Institutet
Browse
- No file added yet -

Biomarkers for early detection and prognostic prediction of hepatocellular carcinoma

Download (2.86 MB)
thesis
posted on 2024-09-19, 07:23 authored by Shifeng LianShifeng Lian

Hepatocellular carcinoma (HCC) is a highly lethal cancer, with over 40% of cases associated with chronic hepatitis B virus (HBV) infection. Early detection of HCC in high-risk HBV- infected populations can greatly improve patient survival; however, effective biomarkers for early detection are currently unavailable. Moreover, reliable biomarkers are needed to accurately predict patient prognosis due to the high heterogeneity of HCC. Liquid biopsy, through the analysis of circulating cell-free DNA (ccfDNA) released by tumor cells, offers a promising non-invasive approach for identifying HCC biomarkers.

This thesis aimed to identify effective ccfDNA biomarkers for HCC, based on a population- based liver cancer screening cohort with a 7-year follow-up of 2,893 HBV-infected individuals. Over 400 samples were analysed, including 270 repeated samples collected before clinical diagnosis (pre-HCC) and non-HCC control samples randomly selected from the screening cohort, as well as 67 samples collected at the time of diagnosis from hospital HCC patients. We developed a novel bilateral single-strand sequencing method to analyse ccfDNA characteristics, including telomere and end sequences (Study I), genome-wide fragmentation (Study II), and tumor content (Study III).

In Study I, we found that the short (25-60 nucleotides) single-stranded telomere G-tail DNA, but not double-stranded telomere DNA, was nearly 20-fold higher in HCC patients compared to non-HCC controls. We developed the telomere and end sequence in circulation (Telecon) model using samples from hospital HCC patients. The Telecon model perfectly distinguished HCC cases from controls, with an area under the curve (AUC) of 1.000 and 95% confidence interval (CI) of 0.998-1.000. Its performance in detecting pre-HCC samples improved as the time of diagnosis approached, with sensitivities increasing from 4.2% (> 4 years before diagnosis) to 59.1% (within 1 year before diagnosis) at 98.0% specificity. In the HBV-infected population, the model had an estimated positive predictive value of 10.2%. A high Telecon score was associated with shorter survival among HCC patients, with a hazard ratio (HR) of 3.22 (95% CI: 1.49-7.00).

In Study II, we observed that HCC patients exhibited greater variation in ccfDNA fragmentation patterns across the genome, particularly on chromosomes 1q and 8q. A fragmentation model was developed using data from hospital HCC patients. This model excellently distinguished HCC patients from controls (AUC: 0.999; 95% CI: 0.997-1.000). The ccfDNA fragmentation score showed increasing performance in detecting pre-HCC samples, with sensitivities rising from 8.3% (> 4 years before diagnosis) to 36.4% (within 1 year before diagnosis) at 88.0% specificity. The positive predictive value was estimated at 1.1% in the HBV-infected population. A high ccfDNA fragmentation score was positively correlated with the Barcelona Clinic Liver Cancer (BCLC) stage and poorer survival (HR: 2.41; 95% CI: 1.13- 5.20).

In Study III, the copy number-based ichorCNA algorithm was used to estimate tumor content in ccfDNA. We observed that tumor content decreased significantly after surgery. In non-HCC control samples, the specificity was 97.8%, with a mean tumor content of 0.011. In pre-HCC samples, the tumor content showed increasing sensitivities from 4.0% (≥ 4 years before diagnosis) to 22.7% (within 1 year before diagnosis), as mean tumor content increased from 0.014 to 0.026. In HCC samples, the sensitivity further increased to 30.4%, 81.8%, and 95.5% for patients at BCLC stages A, B, and C, respectively. HCC patients with high tumor content had poorer survival, particularly those at BCLC stage C (HR: 12.35; 95% CI: 1.42-107.90).

In summary, all three methods-Telecon, genome-wide fragmentation, and tumor content-were effective in detecting HCC at diagnosis and predicting patient prognosis. Telecon, which measures short telomere G-tail, demonstrated the best performance in detecting samples before HCC diagnosis. However, detecting HCC prior to diagnosis remains challenging, and future studies that combine multiple features may improve accuracy.

List of scientific papers

I. Shifeng Lian*, Chenyu Lu*, Fugui Li*, Xia Yu, Limei Ai, Miao Yu, Biaohua Wu, Kuangrong Wei, Wenjing Zhou, Yulong Xie, Yun Du, Wen Quan, Panpan Wang, Li Deng, Zhiheng Liang, Xuejun Liang, Jiyun Zhan, Yong Yuan, Ellen T. Chang, Fang Fang, Zhiwei Liu, Mingfang Ji, Zongli Zheng. Early Detection and Disease Monitoring of Hepatocellular Carcinoma Using Circulating Telomere DNA. (*co-first authors) [Manuscript]

II. Shifeng Lian, Chenyu Lu, Fugui Li, Xia Yu, Limei Ai, Biaohua Wu, Xueyi Gong, Wenjing Zhou, Yulong Xie, Yun Du, Wen Quan, Panpan Wang, Li Deng, Xuejun Liang, Jiyun Zhan, Yong Yuan, Fang Fang, Zhiwei Liu, Mingfang Ji, Zongli Zheng. Circulating DNA genome-wide fragmentation in early detection and disease monitoring of hepatocellular carcinoma. iScience. 2024;27:109701. https://doi.org/10.1016/j.isci.2024.109701

III. Shifeng Lian, Chenyu Lu, Fugui Li, Xia Yu, Limei Ai, Biaohua Wu, Xueyi Gong, Wenjing Zhou, Xuejun Liang, Jiyun Zhan, Yong Yuan, Fang Fang, Zhiwei Liu, Mingfang Ji, Zongli Zheng. Monitoring hepatocellular carcinoma using tumor content in circulating cell-free DNA. Clinical Cancer Research. 2024;30:2772- 9. https://doi.org/10.1158/1078-0432.CCR-23-3449

History

Defence date

2024-10-15

Department

  • Institute of Environmental Medicine

Publisher/Institution

Karolinska Institutet

Main supervisor

Zongli Zheng

Co-supervisors

Fang Fang

Publication year

2024

Thesis type

  • Doctoral thesis

ISBN

978-91-8017-751-1

Number of pages

56

Number of supporting papers

3

Language

  • eng

Author name in thesis

Lian, Shifeng

Original department name

Institute of Environmental Medicine

Place of publication

Stockholm

Usage metrics

    Theses

    Categories

    No categories selected

    Keywords

    Exports

    RefWorks
    BibTeX
    Ref. manager
    Endnote
    DataCite
    NLM
    DC