Journal
EBIOMEDICINE
Volume 30, Issue -, Pages 120-128Publisher
ELSEVIER
DOI: 10.1016/j.ebiom.2018.03.009
Keywords
Lung cancer; Machine learning; Urinary biomarkers
Funding
- Beijing Proteome Research Center
- Tianjin Baodi Hospital
- Ministry of Science and Technology of China [2014DFB30010]
Ask authors/readers for more resources
Development of noninvasive, reliable biomarkers for lung cancer diagnosis has many clinical benefits knowing that most of lung cancer patients are diagnosed at the late stage. For this purpose, we conducted proteomic analyses of 231 human urine samples in healthy individuals (n = 33), benign pulmonary diseases (n = 40), lung cancer (n = 33), bladder cancer (n = 17), cervical cancer (n = 25), colorectal cancer (n = 22), esophageal cancer (n = 14), and gastric cancer (n = 47) patients collected from multiple medical centers. By random forest modeling, we nominated a list of urine proteins that could separate lung cancers from other cases. With a feature selection algorithm, we selected a panel of five urinary biomarkers (FTL: Ferritin light chain; MAPK1IP1L: Mitogen-Activated Protein Kinase 1 Interacting Protein 1 Like; FGB: Fibrinogen Beta Chain; RAB33B: RAB33B, Member RAS Oncogene Family; RAB15: RAB15, Member RAS Oncogene Family) and established a combinatorial model that can correctly classify the majority of lung cancer cases both in the training set (n = 46) and the test sets (n = 14-47 per set) with an AUC ranging from 0.8747 to 0.9853. A combination of five urinary biomarkers not only discriminates lung cancer patients from control groups but also differentiates lung cancer from other common tumors. The biomarker panel and the predictive model, when validated by more samples in a multi-center setting, may be used as an auxiliary diagnostic tool along with imaging technology for lung cancer detection. (C) 2018 Published by Elsevier B.V.
Authors
I am an author on this paper
Click your name to claim this paper and add it to your profile.
Reviews
Recommended
No Data Available