Kato Mamoru, Nishino Jo, Nagai Momoko, Rokutan Hirofumi, Narushima Daichi, Ono Hanako, Hasegawa Takanori, Imoto Seiya, Matsui Shigeyuki, Tsunoda Tatsuhiko, Shibata Tatsuhiro
Division of Bioinformatics, Research Institute, National Cancer Center Japan, Tokyo, Japan.
CREST, JST, Tokyo, Japan.
Hum Genomics. 2025 Apr 12;19(1):39. doi: 10.1186/s40246-025-00744-7.
Cancer prognosis markers are useful for treatment decisions; however, the omics-level landscape is not well understood across multiple cancer types. Pan-Cancer Analysis of Whole Genomes (PCAWG) provides unprecedented access to various types of molecular data, ranging from typical DNA mutations and RNA expressions to more deeply analyzed or whole-genomic features, such as HLA haplotypes and structural variations. We analyzed the PCAWG data of 13 cancer types from 1,514 patients to identify prognosis markers belonging to 17 molecular features in the survival analysis based on the Cox and Lasso regression methods. We found that germline features including HLA haplotypes, neoantigens, and the number of structural variations were associated with overall survival; however, mutational signatures were not. Measuring a few markers provided a sufficient prognostic performance evaluated by c-index for each cancer type. DNA markers demonstrated better or comparable prognostic performance compared to RNA markers in some cancer types. "Universal" markers strongly associated with overall survival across cancer types were not identified. These findings will give insights into the clinical implementation of prognosis markers.
癌症预后标志物对治疗决策很有用；然而，跨多种癌症类型的组学水平情况尚未得到很好的理解。全基因组泛癌分析（PCAWG）提供了前所未有的机会来获取各种类型的分子数据，从典型的DNA突变和RNA表达到更深入分析的或全基因组特征，如HLA单倍型和结构变异。我们分析了来自1514名患者的13种癌症类型的PCAWG数据，以便在基于Cox和Lasso回归方法的生存分析中识别属于17种分子特征的预后标志物。我们发现，包括HLA单倍型、新抗原和结构变异数量在内的种系特征与总生存期相关；然而，突变特征则不然。测量少数标志物对每种癌症类型通过c指数评估提供了足够的预后性能。在某些癌症类型中，DNA标志物与RNA标志物相比显示出更好或相当的预后性能。未发现与跨癌症类型的总生存期密切相关的“通用”标志物。这些发现将为预后标志物的临床应用提供见解。
