Systematic Characterization of Splicing Dysregulation in Pan Solid Tumor Transcriptome

Abstract Splicing dysregulation arising from spliceosomal mutations contributes to disease progression and treatment resistance, mostly in hematologic malignancy. Whereas spliceosomal mutations are less common in solid tumors, splicing disorders are pervasive and proven to promote tumorigenesis. However, there is a lack of systematic understanding of the overall splicing dysregulation patterns and how widespread different patterns occur within or across solid tumor lineage. To address these questions, a computational method called SMNPLS (Sparse Multi‐Network Regularized Partial Least Squares) is developed to uncover the pan‐cancer splicing dysregulation landscape by extracting joint modular patterns from paired matrices of splicing factors (SFs) expressions and alternative splicing events (ASEs). Six unique patterns illustrated by ASE‐SF co‐modules are summarized, which involve 1,570 ASEs and altered expression of 170 SFs, covering 40% of TCGA solid tumors. Cross‐cancer commonalities of splicing dysregulation are observed among digestive system neoplasms, renal‐associated tumors, and urogenital tumors. By contrast, brain tumors demonstrate a distinct splicing pattern with the highest ASE‐SF correlation. In addition, some new splicing regulatory relationships are identified that are potentially oncogenic. Overall, the study characterizes the full spectrum of splicing dysregulation patterns, indicating the similarity and specificity of splicing‐derived pathogenesis across 31 human solid tumors.