Original Article
Proteomics score: a potential biomarker for the prediction of prognosis in non-small cell lung cancer
Abstract
Background: Biomarkers based on quantitative genomics features are related to clinical prognosis in various cancer types. However, the association between proteomics and prognosis in non-small cell lung cancer (NSCLC) is unclear. Here, we developed a proteomics score for the prediction of prognosis in patients with NSCLC undergoing partial pneumonectomy.
Methods: In total, 693 patients with NSCLC with reverse-phase protein array data from The Cancer Genome Atlas were randomly divided into discovery (n=346) and validation (n=347) cohorts. The least absolute shrinkage and selection operator algorithm (LASSO) was used to select the optimal features and build a proteomics score in the discovery set. Additionally, the performance of the proteomics nomogram was estimated using its calibration and time-dependent receiver operator characteristic (ROC) curves. Selection genomics were analyzed via bioinformation.
Results: Using the LASSO model, we established a novel classifier based on 15 features. The proteomics score was significantly associated with overall survival (OS; both P<0.0001) and disease-free survival (DFS; both P<0.0001) in the discovery and validation cohorts. Additionally, the proteomics nomogram showed good discrimination calibration and precise prediction in the two cohorts. Bioinformation revealed that the selection genomics were enriched in negative regulation of immune system processes using gene ontology (GO) and pathways in cancer with the Kyoto Encyclopedia of Genes and Genomes (KEGG).
Conclusions: The proposed proteomics score and nomogram showed excellent performance for the estimation of OS and DFS, which may help clinicians better identify patients with NSCLC who can benefit from surgery.
Methods: In total, 693 patients with NSCLC with reverse-phase protein array data from The Cancer Genome Atlas were randomly divided into discovery (n=346) and validation (n=347) cohorts. The least absolute shrinkage and selection operator algorithm (LASSO) was used to select the optimal features and build a proteomics score in the discovery set. Additionally, the performance of the proteomics nomogram was estimated using its calibration and time-dependent receiver operator characteristic (ROC) curves. Selection genomics were analyzed via bioinformation.
Results: Using the LASSO model, we established a novel classifier based on 15 features. The proteomics score was significantly associated with overall survival (OS; both P<0.0001) and disease-free survival (DFS; both P<0.0001) in the discovery and validation cohorts. Additionally, the proteomics nomogram showed good discrimination calibration and precise prediction in the two cohorts. Bioinformation revealed that the selection genomics were enriched in negative regulation of immune system processes using gene ontology (GO) and pathways in cancer with the Kyoto Encyclopedia of Genes and Genomes (KEGG).
Conclusions: The proposed proteomics score and nomogram showed excellent performance for the estimation of OS and DFS, which may help clinicians better identify patients with NSCLC who can benefit from surgery.
