Application of Bioinformatics in the Revelation of NSCLC Biomarkers and Potential Targeted Drug Therapies

Lung Cancer is the leading cause of cancer-related death in the United States; yet despite this fact, common misconceptions hinder progress towards a better understanding of this fatal carcinoma. It is divided into two crudely differentiated sub-types which are distinguished by the appearance of their respective cells under a microscope: Small-Cell (SCLC) and Non-Small-Cell Lung Carcinomas (NSCLC). In order to find possible genetic perturbations causal for or associated with these diseases, scientists habitually utilize the voluminous capacity of microarray technology, which harnesses the process by which singlestranded DNA, derived from mRNA of differentially expressed genes, hybridized to their complementary probes immobilized on a chip. This genome-wide data is provided to bioinformatics laboratories for analysis, the locus of this investigation. We started with expression microarray data of Adenocarcinoma, Large Cell Carcinoma and Bronchioalveolar Carcinoma (sub-types of NSCLC) tissue from 149 patients. After performing Quality Control Checks to ensure that the data was entirely free of experimental biases, such as the Batch Effect, we proceeded to conduct differential analysis testing on 120 samples from patients with Adenocarcinoma, Large-Cell Carcinoma and Bronchioalveolar Carcinoma: sub-types of NSCLC. Data from these sub-types were put through the Significance Analysis of Microarrays (SAM) Test, to obtain a list of genes that were more significantly expressed in a certain NSCLC sub-types in comparison to its counterparts. The False Discovery Rate (FDR) of significant genes was set to 5% overall or average local if needed. We also conducted a 0% FDRSAM Test in order to identify potential NSCLC Biomarkers. We plugged the filtered 5% FDR differentiallyexpressed gene set into the GRANITE tool which draws Gene Relational Networks and identifies enriched pathways in a set of genes. After identifying these pathways, the most enriched genes in these pathways were plugged into the DrugBank Database in order to isolate specific drugs that would stall a gene, inhibit a rogue pathway or potentially thwart the cancer itself. The resulting genes in the 0% FDR data set could be used in the diagnosis of lung cancer as potential biomarkers of the disease. Hopefully, these beneficial results will pave the way for potential personalized targeted drug therapies to Lung Carcinomas. Introduction Lung Cancer is the leading cause of cancer-related death in the Application of Bioinformatics in the Revelation of NSCLC Biomarkers and Potential Targeted Drug Therapies Mukund Subramanian1*, Jahangheer Shaik2, Rakesh Nagarajan2, and Theodore Waterson3 Student1, Teacher3: Ladue Horton Watkins High School, Saint Louis, MO 63124 Mentor2: Center for Biomedical Informatics, Department of Pathology and Immunology, Washington University, Saint Louis, MO 63124 *Corresponding author: [email protected] INTERNSHIP ARTICLE United States1; in fact the vicious carcinoma claimed the lives of a stunning 158,683 out of 203,536 lung cancer patients in 2007. The scale of lung cancer death is so colossal, that in 2009, a greater number of lung-cancer-diagnosed patients died than those who died of breast, colon, pancreas and prostate cancers combined.9 It is projected that, by 2015, the worldwide number of Lung Cancer deaths alone will amount to a staggering 1,676,000.2 The survival rate for patients with lung cancer is a mere 15% in 5 years: the lowest survival rate of any other cancer.1 Yet current methods used to “treat” lung carcinomas, specifically chemotherapy, have been largely fruitless in their attempts to extend the lives of lung-cancer diagnosed patients. A recent study demonstrated that chemotherapy only contributes to the 5-year survival of 2.3% of all cancer cases.20 In skin cancer research, another study showed that vemurafenib,21 a drug that specifically targets a gene mutation rather than an entire pathway (chemotherapy’s ineffective method), has proven to be 8 times more effective than chemotherapy in shrinking tumors in patients. It seems that the identification of targeted therapies such as this one will lead to a new dawn in treatment. However, in the United States, lung cancer receives just $1,200 of federal funding per death, while breast cancer receives more than $27,000 per death, followed by $14,000 for prostate cancer and $6,500 for colon cancer.22 This lack of funding leads to a subsequent deprivation of research, which has starved the carcinoma of potential targeted therapies. These stunning figures provided the motivation for us to pursue research in this cutting-edge field of oncology. To begin research however, it was imperative that we truly understood Lung Cancer; contrary to common misinterpretations, lung cancer is actually a heterogeneous assortment of tumors of the lung, bronchus, alveoli and pleura.5 It is divided into two crudely differentiated sub-types known as Small-Cell and NonSmall-Cell Lung Carcinoma, indicative of what these respective cells look like under the scrutiny of a microscope. Of these Carcinomas, Non-Small Cell Lung Carcinomas, the focus of this study, represent the overwhelming majority of diagnoses at approximately 80.4%.4 Non-Small Cell Lung Carcinoma (NSCLC) is split into even more definitive sub-types: Adenocarcinoma, the most common form of all lung cancers, Squamous Cell Lung Carcinoma, Large-Cell Lung Carcinoma, and Bronchioalveolar Carcinoma, which, in this study, is considered to have its own explicit NSCLC Histology. These sub-types are widely accepted by the scientific community and provide a basis for oncologists to diagnose and treat with greater accuracy and specificity, due to differences in tumor appearance, location and formation. The need for better analysis of tumors on a genome-wide