Abstract: Objective To study the differential expression of serum low molecular weight proteins in Parkinson's disease(PD) patients and normal subjects. Methods Serum samples from 35 idiopathic PD patients and 35 normal subjects were selected. Serum proteins were captured by weak cationic exchange(WCX) magnetic beads. Molecular weight of the proteins in beads-binding fraction was detected by MALDI-TOF-MS. Differential expression molecules in PD patients and normal subject were screened by statistics. A classification model was constructed by bioinformatics tools like Supervised Neural Network (SNN), and was validated by using 25 newly recruited samples. Results A total of 8 discriminating M/Z peaks related to PD were identified(P<0.05, nonparametric test, Z: -4.458~-3.059). The classification model based on SNN generated a separation between PD and healthy controls. The correct rate was 81.4% for training set, and was 70.0% for 25 newly recruited samples. Conclusion Protein expression in serum of PD patients is different from the normal controls. Serum proteomics data combined with bioinformatics approaches may contribute to the diagnosis of PD.

Key words: Parkinson's disease, proteomics, biomarkers, diagnosis