Multi-omics biomarkers in early gastric cancer screening: Translating discovery evidence to routine screening implementation

Early detection of gastric cancer and precancerous lesions is critical for reducing mortality. Gastroscopy remains the gold standard but is invasive and has limited uptake. Liquid biopsy has enabled discovery of multi‑omics biomarkers across genomics, transcriptomics, proteomics, and metabolomics, with promising diagnostic performance. However, their value for population‑based screening remains unclear. This review summarizes multi‑omics biomarkers across gastric carcinogenesis and assesses their potential for population‑level gastric cancer control.

Introduction

Gastric cancer is among the most common and lethal cancers, especially in East Asia. Five‑year survival exceeds 70% for early‑stage disease but falls below 10% for advanced disease. Gastric carcinogenesis is a multistep process, and identifying precancerous lesions offers opportunities for prevention. Endoscopy is resource‑intensive and has limited adherence, creating a need for non‑invasive pre‑screening tools. Liquid biopsy can detect ctDNA, miRNA, cfRNA, methylation markers, proteins, exosomal cargo, metabolites, and cfDNA fragmentation features. Most multi‑omics research remains in discovery or early validation.

Stage‑Specific Biomarkers

• H. pylori infection and chronic inflammation: Circulating miRNAs (miR‑223, miR‑584, miR‑1290, and others) are associated with H. pylori status.

• Atrophy and intestinal metaplasia: Exosomal miRNAs (miR‑122‑5p, miR‑3591‑3p, miR‑122‑3p) are differentially expressed. Plasma protein panels (ARG1, CA2, HPT, MAN2A1, LBP) discriminate atrophy and intestinal metaplasia. Methylated RPRM cfDNA positivity increases with lesion progression.

• Dysplasia and early gastric cancer: Tissue miRNA changes include downregulation of miR‑26a, miR‑375, and miR‑145, and upregulation of miR‑601, miR‑107, and miR‑18a. Plasma protein markers include GSTP1, CSRP1, and LY6G6F. RPRM methylation is detectable in plasma cfDNA. Extracellular vesicle DNA 5hmC models achieve AUC >0.96.

Liquid Biopsy Biomarkers

• cfDNA mutations: TP53, KRAS, PIK3CA, and APC are commonly detected, but low ctDNA shedding in early disease limits sensitivity.

• cfDNA methylation: mSEPT9 and mRNF180 are validated markers. Genome‑wide profiling has identified 21 gastric cancer‑specific methylated regions with AUC >0.8 and retains tissue‑of‑origin information.

• cfDNA fragmentomics: Fragment size, end motifs, and breakpoints yield multi‑feature models with AUC >0.93 for stage I–II gastric cancer. Machine learning models can reach AUC up to 0.998; the ELSM model achieves AUC 0.922.

• cfRNA and circulating miRNAs: A five‑miRNA panel (miR‑16, miR‑25, miR‑92a, miR‑451, miR‑486‑5p) achieves AUC 0.925. AI‑driven five‑miRNA signatures reach AUC up to 0.986.

• Exosomal miRNAs: miR‑92a‑3p shows AUC 0.829; the DESTINEX 10‑miRNA signature achieves AUC 96.8%.

• Circulating proteins: Conventional CEA and CA19‑9 have limited sensitivity. Protein panels such as B2M, CFL1, CTSD, and HSP90AB1 achieve AUC 0.792. A five‑protein panel (ARG1, CA2, HPT, MAN2A1, LBP) shows AUROC 0.973–0.995. Olink‑based 13‑protein signatures reach AUC up to 0.998.

• Circulating metabolome: A 10‑metabolite panel achieves sensitivity 0.905; serum lipidomic signatures show efficacy, but inter‑individual variability is high.

Comparative Performance

cfDNA fragmentomics reports sensitivity 79–95%, specificity 90–99%, and AUC 0.94–0.98. cfRNA markers show pooled sensitivity and specificity around 80%. cfDNA mutations show high specificity but limited sensitivity (16–85%). cfDNA methylation assays show sensitivity 65–94%, specificity 84–96%, and AUC 0.79–0.96. Platform requirements and costs vary widely, and prospective head‑to‑head evaluation is needed.

Integrative Value

Multi‑omics markers may expand screening coverage, improve adherence, and support risk stratification. Blood‑based testing allows decentralized sampling and centralized analysis, potentially reducing urban–rural disparities. In population surveys, 60.2% preferred blood‑based testing versus 29.8% for endoscopy. Integration into routine health checks may reduce opportunity costs. Molecular risk stratification could reduce unnecessary endoscopies and optimize resource allocation.

Limitations and Challenges

Sensitivity remains insufficient because early tumor signals are low‑abundance. Most studies are small, retrospective, and prone to selection bias and overfitting. They often focus on diagnostic accuracy rather than clinical endpoints. Pre‑analytical standardization is lacking, limiting reproducibility. No consensus exists on actionable clinical thresholds. Integration with existing screening systems requires comparative effectiveness studies.

Future Directions

Large prospective cohorts in high‑risk countries are needed, with clinical endpoints such as reduced late‑stage incidence and mortality. Multi‑omics tests should serve as complementary triage tools rather than replacements for endoscopy. Single‑cell and spatial multi‑omics, AI, and point‑of‑care platforms may accelerate translation.

Conclusions

Multi‑omics biomarkers offer opportunities for non‑invasive risk assessment across gastric carcinogenesis. They may complement endoscopic screening, but validation, standardization, and clinical implementation remain key challenges. Multi‑omics risk stratification has potential to contribute to precision gastric cancer prevention and control.

Full text

https://www.xiahepublishing.com/2835-3315/CSP-2026-00039

The study was recently published in the Cancer Screening and Prevention.

Cancer Screening and Prevention (CSP) publishes high-quality research and review articles related to cancer screening and prevention. It aims to provide a platform for studies that develop innovative and creative strategies and precise models for screening, early detection, and prevention of various cancers. Studies on the integration of precision cancer prevention multiomics where cancer screening, early detection and prevention regimens can precisely reflect the risk of cancer from dissected genomic and environmental parameters are particularly welcome.

Follow us on X: @xiahepublishing

Follow us on LinkedIn:  Xia & He Publishing Inc.