Inherited thrombocytopenia 2 (THC2) is difficult to diagnose due to the lack of specific clinical characteristics and diagnostic methods. To identify potential plasma protein biomarkers for THC2, we collected the plasma samples from a THC2 family (9 THC2 and 15 non-THC2 members), enriched the medium and low abundant proteins using Proteominer and analyzed the protein profiles using the liquid chromatography-mass spectrometry in data independent acquisition mode. Initially, we detected 784 proteins in the plasma samples of this family and identified 27 up-regulated and 36 down-regulated in the THC2 group compared to the non-THC2 group (|log2 ratio| >1 and p-value <0.05). To improve the predictive power, top eight dysregulated proteins (B7Z2B4, LTF, HP, ERN1, IGHV1-8, A0A0X9V9C4, VH6DJ, and D3JV41) were selected by an area under the curve-based random forest process to construct a clinical model. Multivariate analysis with random forest and support vector machine showed that the prediction model provided high discrimination ability for THC2 diagnosis (AUC: 1.000 and 0.967, respectively). The potential plasma protein biomarkers will be tested in more THC2 patients and other thrombocytopenia patients to further validate their specificity and sensitivity.
Keywords: Data independent acquisition (DIA); inherited thrombocytopenia 2 (THC2); platelets; proteomics.