Abstract
Thiol compounds can serve as markers for the antioxidant and prognostic status of lymphoma, playing a crucial role in early tumor diagnosis. However, their high polarity and lack of chromophores pose challenges for multivariate analysis. This study aims to measure thiols associated proteins and develop novel diagnostic models in serum and cerebrospinal fluid from PCNSL to verify potential association and early warning effectiveness. A highly sensitive and selective method was established for simultaneous determination of 7 different thiols and 340 related proteins based on self-developed mass spectrometry probe, Br-OTPP labeled by UHPLC-HRMS. Furthermore, a novel PCNSL monitoring model was developed based on different those combined with machine learning algorithm. Thiols has good linear relationship with correlation coefficients ≥ 0.9995 and suitable precision with inter-day and intra-day coefficients variation was 2.08-5.49% and 1.58-5.53%. Satisfactory accuracy with recoveries between 85.28 and 97.88% was observed. The limit of detection (S/N = 3) was 0.8-9.0 fmol. Ultimately, this method greatly improves discrimination efficiency of model with accuracy of 98.83%. The proposed method could successfully apply into measurement of thiols associated proteins and development of multi-omics clinical evaluation model for PCNSL patients.