Hui Chen1,2, Lingjun Li2, Ping Qin2, Hanzhen Xiong2, Ruichao Chen2, Minfen Zhang2, Qingping Jiang3. 1. Department of Pathology, School of Basic Medical Science, Guangzhou Medical University, Guangzhou, China. 2. Department of Pathology, Third Affiliated Hospital, Guangzhou Medical University, Guangzhou, China. 3. Department of Pathology, Third Affiliated Hospital, Guangzhou Medical University, Guangzhou, China. Jiangqingping@gzhmu.edu.cn.
Abstract
BACKGROUND: Uterine serous carcinoma (USC) is an aggressive type of endometrial cancer that accounts for up to 40% of endometrial cancer deaths, creating an urgent need for prognostic biomarkers. METHODS: USC RNA-Seq data and corresponding patients' clinical records were obtained from The Cancer Genome Atlas and Genotype-Tissue Expression datasets. Univariate cox, Lasso, and Multivariate cox regression analyses were conducted to forge a prognostic signature. Multivariable and univariable cox regression analysis and ROC curve evaluated the prediction efficiency both in the training and testing sets. RESULTS: We uncovered 1385 genes dysregulated in 110 cases of USC tissue relative to 113 cases of normal uterine tissue. Functional enrichment analysis of these genes revealed the involvement of various cancer-related pathways in USC. A novel 4-gene signature (KRT23, CXCL1, SOX9 and ABCA10) of USC prognosis was finally forged by serial regression analyses. Overall patient survival (OS) and recurrence-free survival (RFS) were significantly lower in the high-risk group relative to the low-risk group in both the training and testing sets. The area under the ROC curve of the 4-gene signature was highest among clinicopathological features in predicting OS and RFS. The 4-gene signature was found to be an independent prognostic indicator in USC and was a superior predictor of OS in early stage of USC. CONCLUSIONS: Our findings highlight the potential of the 4-gene signature as a guide for personalized USC treatment.
BACKGROUND: Uterine serous carcinoma (USC) is an aggressive type of endometrial cancer that accounts for up to 40% of endometrial cancer deaths, creating an urgent need for prognostic biomarkers. METHODS: USC RNA-Seq data and corresponding patients' clinical records were obtained from The Cancer Genome Atlas and Genotype-Tissue Expression datasets. Univariate cox, Lasso, and Multivariate cox regression analyses were conducted to forge a prognostic signature. Multivariable and univariable cox regression analysis and ROC curve evaluated the prediction efficiency both in the training and testing sets. RESULTS: We uncovered 1385 genes dysregulated in 110 cases of USC tissue relative to 113 cases of normal uterine tissue. Functional enrichment analysis of these genes revealed the involvement of various cancer-related pathways in USC. A novel 4-gene signature (KRT23, CXCL1, SOX9 and ABCA10) of USC prognosis was finally forged by serial regression analyses. Overall patient survival (OS) and recurrence-free survival (RFS) were significantly lower in the high-risk group relative to the low-risk group in both the training and testing sets. The area under the ROC curve of the 4-gene signature was highest among clinicopathological features in predicting OS and RFS. The 4-gene signature was found to be an independent prognostic indicator in USC and was a superior predictor of OS in early stage of USC. CONCLUSIONS: Our findings highlight the potential of the 4-gene signature as a guide for personalized USC treatment.
Authors: Amanda Nickles Fader; David Starks; Paola A Gehrig; Angeles Alvarez Secord; Heidi E Frasure; David M O'Malley; Erin R Tuller; Peter G Rose; Laura J Havrilesky; Kathleen N Moore; Warner K Huh; Allison E Axtell; Joseph L Kelley; Kristine M Zanotti Journal: Gynecol Oncol Date: 2009-08-26 Impact factor: 5.482