Identifying Novel Sources of Non-canonical Tumor Antigens via the Hybrid de novo Transcriptome Assembly Pipeline

Abstract

Traditionally, tumor-specific antigens (TSAs) are believed to result solely from mutations occurring within specific cancer types. Peptides resulting from these mutations can be applied as therapeutic agents which engineer the body's immune system so that it can identify and remove tumorigenic cells with greater efficiency. These TSA-based therapeutics are known as neoantigen vaccines. Neoantigen vaccines have been developed for personalized treatment of cancer patients, resulting in the overall increased patient survival rates. However, the discovery process of neoantigen vaccines is considered inefficient, often resulting in delays in production, and subsequently, in the delivery to critical cancer patients. Despite developments of sophisticated algorithms, the discovery process of mutation-based neoantigenic TSAs continues to elude researchers with its rarity. The difficulties in the identification process may be attributed to the dynamic nature of the mutational landscape in each tumor. This dynamic nature makes identifying a viable neoantigenic mutation a possibility only to some cancer patients. Furthermore, the rate at which a tumor mutates also varies tumor-to-tumor and patient-to-patient, making a streamline processing of neoantigen vaccine largely impractical. This Thesis serves as a new frontier towards improving TSAs identification and widening the discoverable landscape by identifying novel classes of tumor-specific antigens arising from alternative, non-mutational sources. The work done in this Thesis utilized open-source bioinformatic tools in conjunct with in-house python scripts. The resulting tool is called the novel Hybrid de novo Transcriptome Assembly Pipeline (“hybrid de novo”). The novel hybrid de novo pipeline investigates non-mutational tumorigenic landscapes in carcinomas, sarcomas, and neoplasms in the lung. These landscapes are compared against the landscapes of a healthy human tissue panel (comprising of various tissues of the body) in order to identify the lung TSAs. The procedure largely analyzes total RNA-seqs of the cancer and healthy tissues to identify the presence of any non-canonical transcripts in the non-canonical transcriptional frames. These non-canonical transcriptional frames, which have previously been of little interest in cancer treatments field, have resurfaced as the prime suspect of potential sources for novel TSA isoforms. Utilizing the novel hybrid de novo pipeline, we were able to identify a total of 20 novel, non-canonical TSAs existing only within the lung cancer patient samples (N = 100), all of which were shared amongst the patients with varying degrees of frequency. We also identified an additional 11 novel isoforms that have high expressions in cancer samples, and low expressions in healthy tissues (tumor-associated antigens – TAAs). Summarily, the application of the novel hybrid de novo pipeline in TSAs and TAAs discoveries, in conjunct to traditional pipeline (mutational-based) will improve the overall chance for neoantigen vaccine discovery and clinical translation. Furthermore, the presence of shared TSAs and TAAs show a significant potential for stratifying neoantigen vaccines in patients with similar tumor-genomic make up. If the process for patient genome profiling can be streamlined, groups of cancer patients whose tumor expresses shared TSAs/TAAs will benefit from ready-to-use ‘generic’ neoantigen vaccines. The concept of a ‘generic’ drug for use in personalized medicine will largely reduce the overall time required from lab-to-patient while maintaining the precision associated with traditional personalized medicine.