Background: Wilson disease (WD) is a genetic disorder of copper metabolism. Early diagnosis of WD is inherently challenging due to the absence of typical symptoms. This study aimed to identify urinary protein biomarkers for WD using targeted and nontargeted mass spectrometry-based approaches.
Methods: Exploratory urinary proteomic research on WD patients was initially conducted and revealed some potential biomarkers (alpha-2-macroglobulin, alpha-1-antitrypsin, complement C3, prothrombin, and complement factor B). A multiple reaction monitoring (MRM) assay was subsequently developed and applied to an independent WD cohort for protein candidate validation. Finally, a Random Forest (RF) model constructed with five proteins was evaluated for its diagnostic capacity.
Results: The linear range of the MRM assay extended from 0.025 ng/L to 155 ng/L and the limit of quantification (LOQ) ranged from 0.0095 ng/L to 9.2308 ng/L. Alpha-2-macroglobulin, alpha-1-antitrypsin, and complement C3 exhibited significant increases (p < 0.05) in WD patients compared to the controls, whereas prothrombin and complement factor B only showed variations in concentration. The physiology reference intervals (RIs) for alpha-2-macroglobulin, alpha-1-antitrypsin, complement C3, prothrombin, and complement factor B were estimated as 0-12.50, 0-123.08, 0-5.20, 0-16.59, 0-4.85 ng/mol Cr, while the pathology RIs were 0-114.86, 0-600.98, 0-12.62, 0-22.16, and 0-10.83 ng/mol Cr, respectively. The RF model demonstrated an area under the curve (AUC) of 0.99 for the training data and 0.83 for the testing data.
Conclusions: Based on the proteomic results, the quantitative method was successfully applied for the validation of protein candidates in WD. Using supervised machine learning, the five-protein panel exhibited excellent accuracy in non-invasive diagnosis of WD.
Keywords: LC-MS/MS; Proteomics; Random Forest model; Urine; Wilson disease.
Copyright © 2024 Elsevier B.V. All rights reserved.