Abstract 170: Survival prediction model with long non-coding RNA profile in lung adenocarcinoma cancer

Abstract

Abstract Long non-coding RNAs (LncRNA) are over 200 nucleotides bases non-protein-transcript RNA. Recently, there is a rapid growing literature about LncRNA. LncRNAs are considered as regulator of mRNA, structure components and precursor to small RNAs. However, the mechanism of LncRNA in tumor progression is still not clear yet. Previous studies has validated several LncRNA are associated with prostate cancer and breast cancer (e.g. MATAT-1, PCA3). Here, we investigate the LncRNA expression profile in lung adenocarcinoma cancer and predict patient survival with cox regression model. We download the dataset GSE30219 on Gene Expression Omnibus, microarray data on Affymetrix U133-plus2 platform. With the LncRNA probes selection pipeline on U133-plus2 investigated by Du et al., total 2,427 probes are associated with LncRNA, and every LncRNA is targeted by at least four probes. It is believed that LncRNAs are subtype specific due to previous studies, so we first defined the differential expression LncRNA probes between Adenocarcinoma and Squamous cancer with student t-test, and selected the top 50 p-value probes. Next, with the profile of these 50 probes and the clinical data, we focus on survival analysis of Adenocarcinoma and cox regression was used to identify the survival related probes in the 50 probes, P-value < 0.05. Finally, the expression profiles of survival related probes were trained with cox regression again to build up a prediction model. With score given by the prediction model, samples are divided into two groups: high-risk and low-risk when score higher than 0 or score under 0, respectively. The Kaplan-Meier curve was used to illustrate the relationship between two groups. By using GSE30219 as training model set, GSE30219 was used as internal-validation and GSE8894 was used as external validation. The result of internal-validation shows significantly different. The difference of two groups on external- validation shows, although not greatly as internal- validation, there is still exists a slightly differences. To sum up, LncRNA profiles in Lung Adenocarcinoma might play a role in survival rate. However, further improvement of prediction model is needed since the risk ratio of each sample should be reported, not only the cox scores. Also the Adenocarcinoma specific LncRNA probes should be investigated to enhance the confidence of prediction model. Citation Format: Wei-An Wang, Liang-Chuan Lai, Mong-Hsun Tsai, Tzu-Pin Liu, Eric Y. Chuang. Survival prediction model with long non-coding RNA profile in lung adenocarcinoma cancer. [abstract]. In: Proceedings of the 106th Annual Meeting of the American Association for Cancer Research; 2015 Apr 18-22; Philadelphia, PA. Philadelphia (PA): AACR; Cancer Res 2015;75(15 Suppl):Abstract nr 170. doi:10.1158/1538-7445.AM2015-170