Identification of Specific MRNA Associated with Lymph Node Metastasis in Pancreatic Cancer Using Predictive Models

• Activity Number: 81 - Contributed Poster Presentations: Section on Statistics in Epidemiology
• Type: Contributed
• Date/Time: Monday, August 3, 2020 : 10:00 AM to 2:00 PM
• Sponsor: Section on Statistics in Genomics and Genetics
• Author(s): Nancy Suralik* and Samiran Sinha and Sanjukta Chakraborty and Gordon Olwell
• Companies: Office of the State Comptroller, Medical Fraud Division and Texas A&M and Medical Physiology, College of Medicine , Texas A&M University and Data Scientist
• Keywords: mRNA; Pancreatic Cancer; Random Forests; Genomic; Coxnet Regression; Logistic Regression

Abstract

Pancreatic cancer is an aggressive cancer with a poor survival outcome of less than 5 years. Early metastasis of this cancer to the lymph nodes is an important prognostic factor, yet the molecular mechanisms regulating lymph node metastasis remain ill defined. To determine the prognostic value and association of specific gene expression with increased lymph node metastasis, we used the RNASeq data from The Cancer Genome Atlas to develop a model that predicts the spread of cancer to nearby lymph nodes and also a model which predicts survival time. A binary transformation of the ajcc_pathologic_n variable and survival time were used as responses, and mRNA variables were used as predictors. Logistic regression, the random forest algorithm, and cox regression were used to narrow the number of important mRNAs. Penalized logistic regression and Coxnet regression identified 12 important mRNAs that regulate the spread of the disease to lymph nodes, and 18 mRNAs that control significant variations in the survival time of the patients. In addition, several molecular pathways that could play a role in increased lymph node metastasis were identified and could be promising therapeutic targets.