Department of Thoracic Surgery, Zhongshan Hospital, Fudan University, Shanghai, People's Republic of China.
Center for Tumor Diagnosis and Therapy, Jinshan Hospital, Fudan University, Shanghai, People's Republic of China.
Life Sci. 2021 Apr 1;270:119081. doi: 10.1016/j.lfs.2021.119081. Epub 2021 Jan 28.
To establish a microsatellite instability (MSI) predictive model in pan-cancer and compare the multi-omics characterization of MSI-related molecular features.
We established a 15-gene signature for predicting MSI status and performed a systematic assessment of MSI-related molecular features including gene and miRNA expression, DNA methylation, and somatic mutation, in approximately 10,000 patients across 30 cancer types from The Cancer Genome Atlas, Gene Expression Omnibus database, and our institution. Then we identified common MSI-associated dysregulated molecular features across six cancers and explored their mutual interfering relationships and the drug sensitivity.
we demonstrated the model's high prediction performance and found the samples with high-MSI were mainly distributed in six cancers: BRCA, COAD, LUAD, LIHC, STAD, and UCEC. We found RPL22L1 was up-regulated in the high-MSI group of 5/6 cancer types. CYP27A1 and RAI2 were down-regulated in 4/6 cancer types. More than 20 miRNAs and 39 DMGs were found up-regulated in MSI-H at least three cancers. We discovered some drugs, including OSI-027 and AZD8055 had a higher sensitivity in the high MSI-score group. Functional enrichment analysis revealed the correlation between MSI score and APM score, HLA score, or glycolysis score. The complicated regulatory mechanism of tumor MSI status in multiple dimensions was explored by an integrated analysis of the correlations among MSI-related genes, miRNAs, methylation, and drug response data.
Our pan-cancer study provides a valuable predictive model and a comprehensive atlas of tumor MSI, which may guide more precise and personalized therapeutic strategies for tumor patients.
建立泛癌种微卫星不稳定性（MSI）预测模型，并比较 MSI 相关分子特征的多组学特征。
我们建立了一个用于预测 MSI 状态的 15 基因特征，并对来自癌症基因组图谱（TCGA）、基因表达综合数据库（GEO）和我们机构的约 10000 名 30 种癌症患者的 MSI 相关分子特征进行了系统评估，包括基因和 miRNA 表达、DNA 甲基化和体细胞突变。然后，我们鉴定了六种癌症中常见的 MSI 相关失调的分子特征，并探索了它们之间的相互干扰关系和药物敏感性。
我们展示了该模型的高预测性能，并发现高 MSI 样本主要分布在六种癌症中：BRCA、COAD、LUAD、LIHC、STAD 和 UCEC。我们发现 RPL22L1 在五种癌症中的高 MSI 组中上调，CYP27A1 和 RAI2 在四种癌症中下调。在至少三种癌症中，发现 20 多个 miRNA 和 39 个 DMG 上调。我们发现一些药物，包括 OSI-027 和 AZD8055，在高 MSI 评分组中具有更高的敏感性。功能富集分析显示了 MSI 评分与 APM 评分、HLA 评分或糖酵解评分之间的相关性。通过综合分析 MSI 相关基因、miRNA、甲基化和药物反应数据之间的相关性，探讨了肿瘤 MSI 状态在多个维度的复杂调控机制。
我们的泛癌研究提供了一个有价值的预测模型和肿瘤 MSI 的综合图谱，这可能为肿瘤患者指导更精确和个性化的治疗策略。
