Alternative splicing immune signature for detecting HBV-related hepatocellular carcinoma in peripheral blood mononuclear cells

Peripheral blood mononuclear cells (PBMCs) exhibit disease-specific alternative splicing changes that may serve as robust biomarkers for hepatocellular carcinoma (HCC). Nearly half of the 800,000 annual HCC cases arise from chronic HBV, which affects 254 million people globally, underscoring the urgent need for reliable, non-invasive diagnostic tools. We conducted a comprehensive meta-analysis of PBMC transcriptomes, identifying reproducible immune-related alternative splicing events (ASEs) altered in HBV-related HCC. We identified shifts in inferred immune cell composition and ASEs associated with HBV-related HCC and tumor stage. Using eight immune-related ASEs, we trained a random forest classifier on 120 samples from six independent studies to distinguish HBV-related HCC from chronic HBV, non-HBV liver disease, and healthy controls. The model showed excellent performance in cross-validation (N = 120, AUC: 0.97), internal validation on withheld test samples (N = 54, AUC: 0.99), and external validation with two independent cohorts (N = 27, AUC: 1.00). These findings support PBMC splicing signatures as promising non-invasive biomarkers for early detection and surveillance of HBV-related HCC.