Pathologic evolution-related Gene Analysis based on both single-cell and bulk transcriptomics in Colorectal Cancer

Jiali Li; Zihang Zeng; Jiarui Chen; Xingyu Liu; Xueping Jiang; Wenjie Sun; Yuan Luo; Jiangbo Ren; Yan Gong; Conghua Xie

doi:10.7150/jca.49262

Pathologic evolution-related Gene Analysis based on both single-cell and bulk transcriptomics in Colorectal Cancer

J Cancer. 2020 Oct 22;11(23):6861-6873. doi: 10.7150/jca.49262. eCollection 2020.

Authors

Jiali Li¹, Zihang Zeng¹, Jiarui Chen¹, Xingyu Liu¹, Xueping Jiang¹, Wenjie Sun¹, Yuan Luo¹, Jiangbo Ren², Yan Gong^{2

3}, Conghua Xie^{1

4

5}

Affiliations

¹ Department of Radiation and Medical Oncology, Zhongnan Hospital of Wuhan University, Wuhan, China.
² Department of Biological Repositories, Zhongnan Hospital of Wuhan University, Wuhan, China.
³ Human Genetics Resource Preservation Center of Hubei Province, Zhongnan Hospital of Wuhan University, Wuhan, China.
⁴ Hubei Key Laboratory of Tumor Biological Behaviors, Zhongnan Hospital of Wuhan University, Wuhan, China.
⁵ Hubei Cancer Clinical Study Center, Zhongnan Hospital of Wuhan University, Wuhan, China.

Abstract

Purpose: The patients diagnosed with colorectal cancer (CRC) are likely to undergo differential outcomes in clinical survival owing to different pathologic stages. However, signatures in association with pathologic evolution and CRC prognosis are not clearly defined. This study aimed to identify pathologic evolution-related genes in CRC based on both single-cell and bulk transcriptomics. Patients and methods: The CRC single-cell transcriptomic dataset (GSE81861, n=590) with clinical information and tumor microenvironmental tissues was collected to identify the pathologic evolution-related genes. The colonic adenocarcinoma and rectum adenocarcinoma transcriptomics from The Cancer Genome Atlas were obtained as the training dataset (n=363) and 5 other CRC transcriptomics cohorts from Gene Expression Omnibus (n=1031) were acquired as validation data. Graph-based clustering analysis algorithm was applied to identify pathologic evolution-related cell populations. Pseudotime analysis was performed to construct the trajectory plot of pathologic evolution and to define hub genes in the evolution process. Cell-type identification by estimating relative subsets of RNA transcripts was then executed to build a novel cell infiltration classifier. The prediction efficacy of this classifier was validated in bulk transcriptomic datasets. Results: Epithelial and T cells were elucidated to be related to the pathologic stages in CRC tissues. Pseudotime analysis and survival analysis indicated that HOXC5, HOXC8 and BMP5 were the marker genes in pathologic evolution process. Our cell infiltration classifier exhibited excellent forecast efficacy in predicting pathologic stages and prognosis of CRC patients. Conclusion: We identified pathologic evolution-related genes in single-cell transcriptomic and proposed a novel specific cell infiltration classifier to forecast the prognosis of CRC patients based on pathologic stage-related hub genes HOXC6, HOXC8 and BMP5.

Keywords: GEO; TCGA; colorectal cancer; pathologic stage; prognosis; single-cell sequencing.