Abstract 7072: Lung cancer risk prediction in Asian populations: A study of immune signatures from the Singapore Longitudinal Ageing Studies

Abstract Background: The immune system plays a crucial role in the development and progression of cancer, particularly during aging. Given the distinct genetic and environmental variations in Asian populations compared to Western cohorts, we aimed to evaluate the predictive significance of immune markers for early lung cancer detection in an Asian population. Methods: Two independent cohorts of elderly Asians from the Singapore Longitudinal Ageing Studies (SLAS I/II, NCT03405675) were investigated for immune markers related to lung cancer. Subjects aged 55 years or older, capable of self-ambulation were recruited in 2003 for SLAS I and 2008 for SLAS II (median follow-up period of 17 and 12 years, respectively). Presence of lung cancer served as the clinical endpoint for analysis. Lung cancer diagnoses (69 in SLAS I and 59 in SLAS II) were confirmed through the Singapore National Cancer Registry. Baseline plasma cytokine levels were measured using a 65-plex Human ProcartaPlex Luminex panel (SLAS I: 43 cases, 70 controls; SLAS II: 11 cases, 11 controls). Feature selection was conducted by bootstrapping the summary of 500 differential expression analyses between cases and controls (P<0.05, t-test). Four machine learning algorithms, Logistic Regression, Random Forest, XGBoost, and Support Vector Machine, were used to train models for the prediction of lung cancer using SLAS I, and validated using SLAS II. Characterization of peripheral blood mononuclear cells were performed using a 24-plex flow cytometry on a subset of SLAS II (8 cases, 8 control). Results: A total of 28 cytokines showed significant differences between cases and controls, with TSLP emerging as the most important in the Random Forest, XGBoost models and Logistic Regression model. Comparing models based solely on demographic factors (age, sex, smoking, and presence of COPD) to models that incorporate demographic and Luminex factors, the mean AUC increased from 0.65 to 0.71. External validation showing an AUC improvement from 0.68 to 0.85. Immuno-profiling using flow cytometry revealed a significantly reduced proportion of γδ T cells and reduced CD57-expression in γδ T cell subsets in lung cancer cases. Conclusions: The cytokine TSLP may be a biomarker for assessing the risk of lung cancer. TSLP and its interaction with γδ T cells and CD57-expressing γδ T cells may play key roles in aging-related lung cancer development. Acknowledgements: The work was supported by the A*STAR Biomedical Engineering Programme (C211318003), Industry Alignment Fund-Industry Collaboration Project (I2201E0014; I2001E0065), A*STAR Joint Council Office (1434M00115), A*STAR GAP Funding (I20AGAG602; I23D1AG048), SIgN Immunomonitoring Platform (BMRC IAF 311006 grant; BMRC transition funds #H16/99/b0/011), and Singapore National Medical Research Council (MOH-000323-00; OFYIRG19may-0007; SYMPHONY; MRC/OFLCG/002-2018; OFLCG18May-0028). Citation Format: Jiangfeng Ye, Jia Meng, Gillianne Geet Yi Lai, Xinru Lim, Denise Goh, Weili Xu, Yang Wu, Li Yen Chong, Daniel Shao-Weng Tan, Darren Wan-Teck Lim, Roger Ho, Joe Yeong. Lung cancer risk prediction in Asian populations: A study of immune signatures from the Singapore Longitudinal Ageing Studies [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2025; Part 1 (Regular Abstracts); 2025 Apr 25-30; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2025;85(8_Suppl_1):Abstract nr 7072.