Identifying peripheral cancer-associated TCR signals for the early-detection of lung cancer.

Abstract

2566 Background: Tumor-associated antigens and neoantigens play a critical role in eliciting anti-tumor immune responses, leading to a significant amplification of tumor-specific T cell clones. Consequently, the detection of tumor-associated immune signaling in peripheral blood is expected as a promising strategy for cancer screening. Notably, due to the amplification effect of the immune response, the identification of tumor-related immune signals demonstrates higher abundance compared to the direct detection of tumor-derived molecules, such as circulating tumor DNA (ctDNA). This increased abundance allows for the extension of the screening time window, underscoring its potential for early cancer detection. The present study aims to identify lung cancer-associated T cell receptor (TCR) signatures, and develops a predictive model for to recognize lung cancer based on TCR repertoire sequencing. Methods: Peripheral blood samples were collected from 2,699 lung cancer patients and 3,360 healthy individuals. The TCR repertoires were profiled using a multiplex-PCR-based sequencing of the TCR-β chains. The frequency of TCR clonotypes were calculated using MiXCR tools by aligning against human TCR-β gene segments. The cancer-enriched TCR sequences were identified by comprehensively considering the distribution and frequency of clonotypes in the comparison between the lung cancers and the healthy controls. We proposed an lung cancer-enriched TCR score (LCS) to evaluate the risk for lung cancer based on the CDR3 sequences alignment. A robust machine learning model was developed integrating multiple TCR repertoire characteristics and LCS. Results: The UMAP clustering based on TCR repertoire features revealed distinct TCR characteristics in lung cancer patients compared to healthy individuals. 3,840 TCR clones were identified to be significantly enriched in lung cancer cases. Among these, the CDR3β 'CATSRDTGGREKLFF' was identified as the most highly enriched clone specific to lung cancer patients. Furthermore, a significantly higher LCS was observed in lung cancers (p < 0.001). To validate the utility of LCS, we applied the LCS measurement to an independent public TCR dataset with 382 lung cancer patients, 195 healthy individuals, and 1,034 COVID-19 cases. The external validation demonstrated that lung cancers show a significantly increased LCS compared with healthy controls (p < 0.001), while no significant difference in LCS between COVID-19 cases and healthy controls (p = 0.25). The detection performance of model using integrated TCR features and LCS demonstrated a sensitivity of 0.96 and a specificity of 0.95. Conclusions: Cancer-related immune signals in peripheral blood can inform the anti-tumor responses. This study identified lung cancer-enriched TCR signatures, highlighting the potential as promising biomarkers for lung cancer detection. Clinical trial information: ChiCTR2200055761 .