HIBIT 2021, Ankara, Turkey, 10 - 11 September 2021
Lung cancer is the most common disease worldwide and one of the deadliest diseases. Over 2.2 million
new diagnoses are made annually, resulting in approximately 82% of deaths (WHO, 2021). Non-small cell lung
cancer (NSCLC), which accounts for approximately 80% of lung cancer types, is the most common subtype. As
for other cancers, there are various treatment options such as chemotherapy for lung cancer; however, survival
rates are very low, and poor prognosis due to relapse or drug-resistant cases is still a major problem.
In the personalized medicine field, immuno-oncology approaches have gained attraction from the
research community. One of the cutting-edge therapies is called chimeric antigen receptor (CAR) T cell therapy.
With this approach, engineered chimeric receptors on cytotoxic T cells can increase binding specificity by
recognizing the 3D structure of proteins on the outer surface of the tumor cell membrane as antigens. Thus,
reprogrammed T cells that carry these so-called “specialized weapons” are expected to target and kill any tumor
cell that carries these proteins on the cell surface. Even though it’s a highly promising strategy, there is
currently no FDA-approved CAR T cell therapy for NSCLC. One of the most important reasons is the low
specificity of existing tumor antigens and their high side effects. Therefore, the determination of new target
antigens and increasing the efforts to expand the candidate pool will open the doors of new immunotherapybased
clinical trials for NSCLC as well.
Expanding the candidate antigen pool is essential and more research efforts need to be channeled
towards searches for new tumor-specific targets. Better bioinformatics workflows that can streamline target
antigen identification and selection process are required. In the last 10 years, single-cell sequencing
technologies have gained tremendous speed and have been the main approach used in more than 1200 studies.
In these studies, more than 17 million individual cells have undergone scRNAseq for various purposes. Of
these, roughly 10% represents various types of tumor cells providing new insights into tumor-specific
transcriptome profiles at the single-cell resolution. Here in this study, we hypothesized that repurposing these
datasets to identify alternative tumor-specific antigens for CAR T cell therapy for NSCLC can be achieved with
appropriate data integration and analytical processing. To achieve this, we developed a target enrichment
scheme that meticulously integrates various scRNAseq datasets, filters, normalizes, and prepares for
differential gene expression analysis between multiple combinations of cell types. This complex analysis is
successful only if cell type identities are correctly determined. Our analysis pipeline utilizes a machine learning
approach, i.e., HieRFIT (Kaymaz et al., 2021 Bioinformatics), for hierarchical cell type classification of tumor
and normal cell types. Thus, better tumor-specific antigens will be selected, and their toxigenic effects will be
foreseen when used in CAR T cell therapy.
Here, we will introduce our analytical approach in detail and share our preliminary results leading to
an effective antigen candidate. We believe that our work will contribute to a paradigm shift in the selection of
candidate antigens for CAR T cell therapy by reprocessing existing single-cell transcriptome data. This datamining
proof-of-concept using single-cell sequencing and metadata will help expand the scope of available
treatments for other malignancies as well.
Key phrases: NSCLC, CAR T cell therapy, Data repurposing, Data integration, Single-cell RNAseq, Cell types,