肺癌生物标志物从基于纳米颗粒的LCMS向酶联免疫吸附试验的转化

Lung cancer is the leading cause of cancer-related mortality in the United States. Newly diagnosed lung cancer patients generally have poor prognoses, in large part due to being diagnosed at later stages of disease. Earlier diagnoses, enabled by more effective screening, are expected to reduce morbidity and mortality. In pursuit of this, we previously reported on an unbiased, multi-omics discovery study to identify blood-based biomarkers for lung cancer that may be developed into a more effective screening test. A machine-learned model trained on nanoparticle-based LCMS measurements could achieve a specificity of 74% at a sensitivity of 85% across all lung cancer stages and sensitivity of 74% on stage-1 lung cancer alone. These results illustrate the power of unbiased LCMS plasma proteomics to identify proteins with high disease discrimination; however, there is limited demonstration of translating findings from nanoparticle-based unbiased LCMS proteomics to immunoassays. Here, we report on the translation of 8 of the most cancer discriminative, plasma proteins identified from our discovery study to enzyme-linked immunosorbent assay (ELISA). On a set of 404 subjects (110 cancer and 294 non-cancer) from the discovery study, a machine-learned model trained on nanoparticle-based LCMS measurements of these 8 proteins achieved an AUROC of 0.93 (95% CI 0.90-0.96) and > 80% specificity at 87.5% sensitivity. The corresponding model trained on ELISA measurements of the same 8 proteins achieved an AUROC of 0.90 (95% CI 0.86-0.94) and > 70% specificity at 87.5% sensitivity. The directionality and magnitude of the fold-changes between cancer and non-cancer subjects were preserved for each of the 8 proteins between the two assays. Statistically significant (adjusted p-value < 0.05) and positive Spearman correlations were also observed between measurements of each of the 8 proteins on these two assays across the 404 subjects. These results demonstrate the feasibility of translating from nanoparticle-based LCMS to immunoassays while preserving lung cancer discriminative signals and set the foundation for the development of an immunoassay-based Lab Developed Test (LDT) for lung cancer detection.