肿瘤相关抗原的高分辨率图谱用于基于自身抗体的肺癌检测

Introduction Non-invasive, blood-based screening for lung cancer is a promising approach to early cancer detection, but is challenged by low concentrations of tumor-derived biomarkers, especially in early-stage disease. Auto-antibodies (AAbs) are a compelling class of biomarkers to address this gap, leveraging the body's amplified humoral immune response to tumor antigens and providing a unique source of signal not directly linked to tumor shedding. However, natural variation in human immune profiles makes identification of true cancer-specific antigens challenging. Methods In this work, we used a Phage Immunoprecipitation and sequencing (PhIP-Seq) method to identify AAbs differentially present in patients with lung cancer. Unlike traditional proteome-wide PhIP-Seq approaches, we used cancer mutation data along with tumor-specific protein expression patterns from GTEx and TCGA databases to narrow the search space to ~4000 proteins with high probability of generating a cancer-specific immune response. By densely tiling these proteins with overlapping 54-mer peptides and requiring overlapping peptides for hit calling, we maximized technical reproducibility while also mapping cancer-specific antigenic regions within proteins to high resolution. Applying this approach to 1,200 samples enabled us to differentiate cancer-specific signals from background antigenicity. Results We screened 400 samples from patients with lung cancer and 800 samples from age- and sex-matched healthy controls. Our hit-calling pipeline identified antigens showing significant AAb signal in plasma from multiple cancer patients and little or no signal in healthy controls, taking into account both technical and biological noise. Based on this pipeline, we identified 90 cancer-specific peptide antigens spanning 68 human proteins, including established lung cancer biomarkers such as p53 and NY-ESO-1 as well as proteins not previously associated with lung cancer autoimmunity. Importantly, even in well-established tumor-associated antigen (TAA) proteins, we identified regions with high AAb positivity in healthy individuals, showing that selecting specific antigenic regions of TAA proteins could be crucial for maximizing signal-to-noise. Moreover, we observed distinct AAb profiles across different patient subsets, indicating that a large panel of antigens may help achieve higher sensitivity in a blood-based cancer detection assay. Conclusions This work represents a significant step forward in mapping the cancer humoral immunome. By performing PhIP-Seq based AAb profiling in the largest cohort of lung cancer patients and matched healthy controls published to date, and enabling deep peptide-level characterization of the auto-antibody response, we maximize confidence in identification of cancer-specific biomarkers and provide a path to a more accurate AAb-based cancer detection test.