The characterization of serum proteomics and metabolomics across the cancer trajectory in chronic hepatitis B‐related liver diseases
Jin Xiao,Hang Liu,Jun Yao,Shuang Yang,Fenglin Shen,KunPeng Bu,Zhenxin Wang,Fan Liu,Ningshao Xia,Quan Yuan,Hong Shu,Yueting Xiong,Xiaohui Liu
DOI: https://doi.org/10.1002/viw.20240031
2024-10-28
View
Abstract:We have established an extensive biomarker bank comprising 3059 proteins and 103 metabolites as potential biomarkers for hepatocellular carcinoma (HCC). By employing targeted proteomics (DeepPRM‐MS) and metabolomics (scheduled‐MRM‐MS), we conducted a screening for promising biomarkers aimed at the early detection of HCC within a cohort of 316 individuals. Through the integration of advanced machine learning algorithms, we successfully identified two distinct sets of biomolecular combinations: Com‐8, which serves as a serum biomarker for the early diagnosis of HCC and demonstrates exceptional diagnostic capabilities, and Com‐12, which effectively predicts the progression from chronic hepatitis B to cirrhosis. These combinations surpass existing single biomarker or single omics approaches in terms of accuracy, significantly addressing limitations related to specificity and sensitivity. Hepatocellular carcinoma (HCC) is a deadly cancer that emerges from a continuous progression of liver cells from normal to abnormal, often following infections by hepatitis B/C viruses (HBV/HCV), liver fibrosis, and liver cirrhosis (LC), ultimately culminating in cancer. However, there is currently limited systematic molecular analysis of biomarkers at different stages of HCC progression using multi‐omics approaches. We carried out an innovative pipeline by utilizing targeted proteomics and metabolomics to identify potential biomarkers for early detection of HCC in 316 participants, including healthy adults and patients diagnosed with HBV, HCV, LC, and HCC from three independent cohorts. We first established a detailed database of candidate biomarkers for HCC containing 3059 proteins and 103 metabolites, and identified pivotal candidates implicated in the progressive trajectory of liver cancers. Through our developed DeepPRM, scheduled multiple reaction monitoring (MRM)‐targeted approach, and machine learning‐based computational pipeline, we identified an eight‐biomolecular‐based combination with an accuracy rate of 91.43% for early diagnosis of HCC, and a 12‐biomolecular‐based combination with an accuracy rate of 80.00% for detecting changes in HBV–LC progression. These two biomarker combinations significantly improved accuracy compared to traditional tumor biomarkers. Our extensive analysis provides valuable proteomic and metabolomic data resources that will contribute to a deeper understanding of liver disease progression and enhance the identification of potential therapeutic targets.
materials science, biomaterials,nanoscience & nanotechnology