Baseline Plasma Proteomic Analysis to Identify Glycoproteins for Prediction of Radiation Induced Lung Toxicity in Patients with Non-Small Cell Lung Cancer

J. Wang,X. Xie,F. Peng,P. Stanton,S. Nie,L. Sun,W. Wang,D. M. Lubman,F. M. P. Kong

DOI: https://doi.org/10.1016/j.ijrobp.2015.07.360

2015-01-01

Abstract:Radiation induced lung toxicity (RILT) is one of the most important dose-limiting toxicities for patients with lung cancer receiving thoracic radiation therapy; herein, we sought to identify baseline (prior to radiation therapy start) plasma proteomic markers to predict grade ≥ 2 radiation induced lung toxicity (RILT2). Thirty patients with prospective blood sample collection including 8 patients with RILT2 and 22 patients without RILT were eligible for this analysis. Platelet-poor plasma was obtained within 2 weeks prior to radiation therapy. After depleting 14 high-abundance proteins and filtering proteins less than 3KD, we applied lectin microarrays to investigate the potential differential glycan structures between two groups of plasma, followed by lectin affinity chromatography to enrich the glycoproteins with specific structures. The resulting proteins were digested and then analyzed by LC-MS/MS using a LTQ mass spectrometer. Identified proteins were quantified using a spectral counting-based, label-free quantitative strategy. All MS/MS spectra were searched against the UniProt human protein database Release 2010_11. The search was performed using SEQUEST algorithm version 29 incorporated in Proteome Discoverer software version 1.1. Enzyme-linked immunosorbent assay (ELISA) was performed to independently validate the differential expression of discriminative glycoproteins. Two lectins including concanavalin A (Con A) and Aleuria Aurantia Lectin (AAL) were identified by lectin array and further used to capture glycoproteins for this study from plasma. Finally, a total of 325 proteins were identified and quantified. After excluding proteins that were not detected in at least 50% of patient samples, 4 glycoproteins had significantly different expression levels between patients with RILT2, and those without RILT2, including lower expression of plasma protease C1 inhibitor (P = .001) and haptoglobin (P = .002) and higher expression of serum amyloid P component (P < .001) and Zinc-alpha-2-glycoprotein (P = .003) in RILT2 patients. The expression of these proteins was validated using ELISA method, where interestingly, all of them had biological functions associated with inflammation, fibrosis, or immunomodulation. This study suggests that glycoprotein markers may characterize differentiation status of predisposition to RILT, and identify patients at high-risk for RILT. This study is ongoing in our center to validate these findings in an independent patient population of a larger sample size.