Background: Liver AS has a grave prognosis and our previous study (Chen TW et al. Cancer 2022) showed that the incidence of liver AS is higher in Asia. We aimed to investigate the etiology of liver AS and its molecular characteristics. Methods: We performed whole exome (WES) and transcriptome sequencing (RNA-seq) of AS (liver 10, scalp 5, breast 2, trunk 1, heart 1) and hepatocellular carcinoma (HCC, 9) from National Taiwan University Hospital (NTUH) and re-analyzed WES & RNA-seq of AS from the Count Me In Project (CMI, 46). Somatic mutations and copy number alternation were called by Mutect2 and Sequenza from tumor-normal matched WES. Non-negative matrix factorization was used to determine mutational signature profiles and the contributions of each signature to each sample. Telomere length was estimated using TelSeq. Results: We identified four single-base-substitution (SBS) signatures using somatic mutations from 74 (AS/HCC 65/9) WES data. The dominated signature in liver AS had exceptionally high cosine similarity (93.08%) with COSMIC SBS22 (ver 3.2), which is related to AA exposure. The AA-mutation signature contributed significantly (adjusted p < 0.01) to the DNA mutations in 70% (7/10), 0% (0/9), 7% (3/46), and 56% (5/9) of liver AS (NTUH), nonliver AS (NTUH), CMI AS, and HCC (NTUH), respectively. TP53 (60% vs 22% vs 32%) and ATRX (40% vs 0% vs 11%) mutations were more common in liver than nonliver AS (NTUH) or CMI, while KDR mutation rates were similar (30% vs 44% vs 22%). Among TP53 mutations from liver AS with AA signature, 86% (6/7) showed the typical A:T > T:A mutation caused explicitly by AA-induced DNA damage. In liver vs nonliver AS, the median number of mutations and neoantigen burdens were 6.55 vs 0.98 (p = 0.0076) and 1572 vs 544 (p = 0.095), respectively. Although liver AS and HCC were highly associated with AA-mutation signature, liver AS had significantly longer telomeres in terms of tumor/normal telomere ratio (median 1.27 vs 1.16, p = 0.0009) but a significantly lower telomerase enzyme activity (p = 0.016). We did not find an association between ATRX mutation, or its mRNA expression level, with the extended telomere length in liver AS. Compared to nonliver AS, the activity of the telomere replication molecule BUB1 was significantly higher in liver AS (p = 0.0029). Conclusions: Our study is the first to connect AA exposure to the etiology of liver AS. The high mutation rate and distinct telomere maintenance mechanism of liver AS provides insights into future treatments.