High throughput immunoproteomics for cancer biomarker discovery

Background: Early detection of lung cancer (LC) remains a critical unmet need. While computed tomography (CT) screening saves lives by detecting cancer at early stages, its utility is limited by high false positive rates and insufficient sensitivity. These limitations lead to unnecessary surgeries and missed malignancies, particularly in individuals presenting with indeterminate pulmonary nodules (IPNs). Objective: Our overarching goal is to develop circulating biochemical biomarkers that improve the specificity of CT screening for LC by differentiating malignant from benign IPNs. Methods: We applied a high-throughput systems immunoproteomics strategy to discover serum biomarkers able to discriminate between malignant and benign IPNs. This integrated approach profiles three classes of circulating biomarkers: autoantibodies, anti-microbial antibodies, and serum proteins. Results: In the discovery phase, we profiled IgG and IgA autoantibodies using Nucleic Acid Programmable Protein Array (NAPPA) against 13,330 full-length human proteins, along with microbial antibodies against 8,820 microbial antigens. These analyses were conducted using serum from 144 lung cancer cases and 143 benign controls from Vanderbilt University Medical Center. Antibodies with significant enrichment in cases (odds ratio p < 0.05 in the top decile) were prioritized, yielding 112 autoantibodies and 70 microbial antibodies associated with malignancy, as well as 50 autoantibodies and 230 microbial antibodies associated with benign disease. In the validation phase, we assessed the prioritized antibody candidates in 319 subjects from the Detection of Early Lung Cancer Among Military Personnel (DECAMP-1) cohort using our Multiplexed In-Solution Protein Array (MISPA). In parallel, we quantified 19 well-reported cancer-associated serum proteins across the same samples. A multimodal panel comprising 7 autoantibodies, 4 microbial antibodies, and 4 serum proteins achieved an area under the curve (AUC) of 0.81 in the discovery cohort and 0.74 in independent validation cohorts, showing improved discrimination of malignant versus benign nodules compared to clinical models alone. Additionally, to demonstrate clinical scalability, we confirmed performance of top markers in 46 cases and 230 controls from the German Lung Cancer Screening Intervention Study (LUSI) using the Meso Scale Discovery (MSD) electrochemiluminescence platform. Conclusions: These results demonstrate a fully integrated immunoproteomics pipeline for the discovery, validation, and potential clinical translation of multiplex antibody and protein biomarkers for lung cancer detection.