IntroductionType B1 and B2 thymomas are lymphocyte-rich malignant tumors with few somatic mutations in protein-coding regions of the nuclear genome; nonetheless, non-coding regions remain uncharacterized. Here, we developed a rigorous tumor isolation method from lymphocyte-rich thymoma tissues and identified somatic mutations in non-coding and mitochondrial DNA.MethodsCD205+CD45-pure tumor cells were isolated from fresh-frozen tissues using DEPArray system. Deep whole-genome sequencing was performed, and recurrent somatic alterations in coding, non-coding, and mitochondria regions were systemically identified by computational framework. The mutations were classified according to gene function, cis-regulatory element, and mutational signature.ResultsThe total number of somatic mutations was approximately 80 times higher in non-coding regions than in coding regions in type B1-2 thymomas (1,671.3 vs. 21.1 per case). Coding mutations were identified in epigenetic regulators, DNA repair genes, and some other genes. Nevertheless, 40% of cases exhibited fewer than four mutations in coding regions. A systematic non-coding analysis identified a total of 405.0 mutations per case on cis-regulatory elements, and detected six recurrent mutations: one interferon regulatory factor (IRF8), two E3 ubiquitin ligases (UBR2andRNF213), and three intergenic regions. Mitochondrial heteroplasmy was observed in 90% of cases, with a significant proportion of mutations located in D-loop region. The single-base substitution pattern was signature 12.ConclusionsNumerous non-coding mutations and mitochondrial heteroplasmy were detected in type B1 and B2 thymomas. Given the paucity of coding mutations observed in this disease entity, disruption of the non-coding landscape and mitochondrial heteroplasmic shift may be the primary cause of thymoma.