Plasma Proteomics Enable Differentiation of Lung Adenocarcinoma from Chronic Obstructive Pulmonary Disease (COPD)
2022
Thilo Bracht | Daniel Kleefisch | Karin Schork | Kathrin E. Witzke | Weiqiang Chen | Malte Bayer | Jan Hovanec | Georg Johnen | Swetlana Meier | Yon-Dschun Ko | Thomas Behrens | Thomas Brüning | Jana Fassunke | Reinhard Buettner | Julian Uszkoreit | Michael Adamzik | Martin Eisenacher | Barbara Sitek
Chronic obstructive pulmonary disease (COPD) is a major risk factor for the development of lung adenocarcinoma (AC). AC often develops on underlying COPD; thus, the differentiation of both entities by biomarker is challenging. Although survival of AC patients strongly depends on early diagnosis, a biomarker panel for AC detection and differentiation from COPD is still missing. Plasma samples from 176 patients with AC with or without underlying COPD, COPD patients, and hospital controls were analyzed using mass-spectrometry-based proteomics. We performed univariate statistics and additionally evaluated machine learning algorithms regarding the differentiation of AC vs. COPD and AC with COPD vs. COPD. Univariate statistics revealed significantly regulated proteins that were significantly regulated between the patient groups. Furthermore, random forest classification yielded the best performance for differentiation of AC vs. COPD (area under the curve (AUC) 0.935) and AC with COPD vs. COPD (AUC 0.916). The most influential proteins were identified by permutation feature importance and compared to those identified by univariate testing. We demonstrate the great potential of machine learning for differentiation of highly similar disease entities and present a panel of biomarker candidates that should be considered for the development of a future biomarker panel.
Show more [+] Less [-]AGROVOC Keywords
Bibliographic information
This bibliographic record has been provided by Directory of Open Access Journals
