Abstract 2232: Preliminary results for a novel single extracellular vesicle assay for early lung cancer: The power of co-localized detection of surface biomarkers

Introduction: Screening for lung cancer (LC), the leading cause of cancer deaths, with helical computerized tomography lowers mortality but uptake is poor. Investigations into new approaches such as using circulating tumor cells and circulating tumor DNA for LC detection have soared in the last decade. However, the low abundance of these targets has limited the performance of these approaches as screening tools. We hypothesize that co-localization of biomarkers on the surface of individual extracellular vesicles (EVs), which are shed into the circulation by cancer cells, may lead to development of a blood test for early stage LC. We evaluated the potential of our approach in detecting early stage LC in clinical samples. Methods: EVs were purified from plasma using size-exclusion chromatography and immunoaffinity capture, and biomarkers co-localized on the EV surface were detected with proximity ligation qPCR. We used antibody combinations comprising 1 capture antibody and 2 oligonucleotide-tagged detection antibodies, recognizing 1, 2 or 3 unique biomarkers. We evaluated this approach by testing plasma samples from early stage I/II lung adenocarcinoma (LUAD) patients (15 smokers, 19 non-smokers), late stage III/IV LUAD patients (16 smokers, 18 non-smokers), and healthy donors (34 smokers, 33 non-smokers). Samples were from one vendor, processed using a standardized protocol. LUAD samples were sourced from a cancer research center and healthy samples from a primary care facility. PCR cycle threshold (Ct) values were generated for each combination and data was evaluated using univariate analysis. Results: Combinations recognizing 3 biomarkers were better in detecting all stages of LUAD (AUC=0.83, 95% CI 0.77-0.90), as compared to combinations recognizing 2 biomarkers (AUC=0.71, 95% CI 0.63-0.80) or 1 biomarker (AUC=0.50, 95% CI 0.35-0.55), demonstrating greater accuracy with an increasing number of co-localized biomarkers. In detecting LUAD (all stages) at a specificity of 0.80 (95% CI 0.69-0.88), sensitivity improved as the number of co-localized biomarkers increased from 1 (0.08, 95% CI 0.03-0.18) to 2 (0.60, 95% CI 0.48-0.72) to 3 (0.76, 95% CI 0.65-0.86). In detecting early stage I/II LUAD, the most effective combination used 3 biomarkers (STn, MUC1, CEACAM6) and had a sensitivity of 0.56 (95% CI 0.38-0.73). Conclusions: These preliminary data highlight the potential of detecting biomarkers co-localized on the surface of single EVs as an effective tool f