Birgitte Sandfeld-Paulsen1, Kristine Raaby Jakobsen2, Rikke Bæk3, Birgitte Holst Folkersen4, Torben Riis Rasmussen4, Peter Meldgaard5, Kim Varming3, Malene Møller Jørgensen3, Boe Sandahl Sorensen6. 1. Department of Clinical Biochemistry, Aarhus University Hospital, Aarhus, Denmark. Electronic address: birgne@rm.dk. 2. Department of Clinical Biochemistry, Aarhus University Hospital, Aarhus, Denmark; Department of Biomedicine, Aarhus University, Aarhus, Denmark. 3. Department of Immunology, Aalborg University Hospital, Aalborg, Denmark. 4. Department of Pulmonary Medicine, Aarhus University Hospital, Aarhus, Denmark. 5. Department of Oncology, Aarhus University Hospital, Aarhus, Denmark. 6. Department of Clinical Biochemistry, Aarhus University Hospital, Aarhus, Denmark.
Abstract
INTRODUCTION: Exosomes have been suggested as promising biomarkers in NSCLC because they contain proteins from their originating cells and are readily available in plasma. In this study, we explored the potential of exosome protein profiling in diagnosing lung cancers of all stages and various histological subtypes in patients. METHODS: Plasma was isolated from 581 patients (431 with lung cancer and 150 controls). The extracellular vesicle array was used to phenotype exosomes. The extracellular vesicle array contained 49 antibodies for capturing exosomes. Subsequently, a cocktail of biotin-conjugated CD9, CD81, and CD63 antibodies was used to detect and visualize captured exosomes. Multimarker models were made by combining two or more markers. The optimal multimarker model was evaluated by area under the curve (AUC) and random forests analysis. RESULTS: The markers CD151, CD171, and tetraspanin 8 were the strongest separators of patients with cancer of all histological subtypes versus patients without cancer (CD151: AUC = 0.68, p = 0.0002; CD171: AUC = 0.60, p = 0.0002; and TSPAN8: AUC = 0.60, p = 0.0002). The multimarker models with the largest AUC in the cohort of patients with all lung cancer histological subtypes and in the cohort of patients with adenocarcinoma only covered 10 markers (all cancer: AUC = 0.74 [95% confidence interval: 0.70-0.80]; adenocarcinoma only: AUC = 0.76 [95% confidence interval: 0.70-0.83]). In squamous cell cancer and SCLC, multimarker models did not exceed CD151 as an individual marker in separating patients with cancer from controls. CONCLUSION: We have demonstrated exosome protein profiling to be a promising diagnostic tool in lung cancer independently of stage and histological subtype. Multimarker models could make a fair separation of patients, demonstrating the perspectives of exosome protein profiling as a biomarker.
INTRODUCTION: Exosomes have been suggested as promising biomarkers in NSCLC because they contain proteins from their originating cells and are readily available in plasma. In this study, we explored the potential of exosome protein profiling in diagnosing lung cancers of all stages and various histological subtypes in patients. METHODS: Plasma was isolated from 581 patients (431 with lung cancer and 150 controls). The extracellular vesicle array was used to phenotype exosomes. The extracellular vesicle array contained 49 antibodies for capturing exosomes. Subsequently, a cocktail of biotin-conjugated CD9, CD81, and CD63 antibodies was used to detect and visualize captured exosomes. Multimarker models were made by combining two or more markers. The optimal multimarker model was evaluated by area under the curve (AUC) and random forests analysis. RESULTS: The markers CD151, CD171, and tetraspanin 8 were the strongest separators of patients with cancer of all histological subtypes versus patients without cancer (CD151: AUC = 0.68, p = 0.0002; CD171: AUC = 0.60, p = 0.0002; and TSPAN8: AUC = 0.60, p = 0.0002). The multimarker models with the largest AUC in the cohort of patients with all lung cancer histological subtypes and in the cohort of patients with adenocarcinoma only covered 10 markers (all cancer: AUC = 0.74 [95% confidence interval: 0.70-0.80]; adenocarcinoma only: AUC = 0.76 [95% confidence interval: 0.70-0.83]). In squamous cell cancer and SCLC, multimarker models did not exceed CD151 as an individual marker in separating patients with cancer from controls. CONCLUSION: We have demonstrated exosome protein profiling to be a promising diagnostic tool in lung cancer independently of stage and histological subtype. Multimarker models could make a fair separation of patients, demonstrating the perspectives of exosome protein profiling as a biomarker.
Authors: Safinur Atay; Daniel W Wilkey; Mohammed Milhem; Michael Merchant; Andrew K Godwin Journal: Mol Cell Proteomics Date: 2017-12-14 Impact factor: 5.911