Abstract
INTRODUCTION: The altered expression of genes encoding histone acetyltransferases (HATs) and histone deacetylases (HDACs) has been implicated in the tumorigenesis and progression of various solid tumors. However, systematic characterization of the transcriptomic landscape and clinical relevance of HATs and HDACs in pan-cancer contexts remains lacking. METHODS: Transcriptome and clinical data of 9,483 patients across 31 tumor types from The Cancer Genome Atlas were collected for systematic pan-cancer analysis. Additional glioma-specific datasets (Chinese Glioma Genome Atlas, GlioVis, GSE43378, and GSE182109) were also collected to validate the transcriptional characteristics of HATs and HDACs in gliomas. Consensus clustering analysis was applied to identify distinct expression patterns of HATs and HDACs. RESULTS: Based on the transcriptomic data of 25 genes encoding 9 HATs and 16 HDACs, we identified five major subtypes across 31 cancer types (AC-I to AC-V). Notably, the AC-V subtype comprised over 95% of glioma patients, suggesting glioma patients exhibited distinct expression patterns of histone acetylation-modifying enzymes compared to patients with other solid tumors. Therefore, we re-conducted the consensus clustering analysis specifically within the context of gliomas and identified five subtypes, denoted "AC-GI" to "AC-GV", which were characterized by differences in HATs/HDACs expression patterns, biological and immune status, genetic alterations, and clinical outcomes. The AC-GII patients exhibited the best prognosis and were sensitive to temozolomide, while AC-GV patients had the poorest prognosis and the lowest sensitivity to temozolomide among all subtypes. Moreover, based on the Connectivity Map database analysis and experimental verification, we identified several pan-HDAC inhibitors that could serve as sensitizers for temozolomide therapy in AC-GV patients, such as panobinostat and scriptaid. Considering the distinctive clinical characteristics of patients with AC-GII and AC-GV, we constructed the "ACG score" model capable of effectively recognizing patients with these subtypes and predicting patient prognosis. CONCLUSION: Herein, we established novel biologically and clinically relevant molecular classifications for pan-solid tumors and gliomas based on transcriptional expression profiles of HATs and HDACs. Moreover, the ACG score model, calculated by the transcriptional expression of 29 genes, was not only an independent prognostic factor for glioma patients, but can also provide valuable references for promoting more effective therapeutic strategies.