Serum profiling by mass spectrometry combined with bioinformatics for the biomarkers discovery in diffuse large B-cell lymphoma

Wenhong Xu, Yue Hu, Xuexin He, Jun Li, Tao Pan, Hai Liu, Xianguo Wu, Hong He, Weiting Ge, Jiekai Yu, Qichun Wei, Shu Zheng, Suzhan Zhang, Yiding Chen

The aim of this study was to identify potential serum biomarkers of diffuse large B-cell lymphoma (DLBCL) and to detect DLBCL therapy response biomarkers. DLBCL serum proteomic analysis was performed using the CM10 ProteinChip mass spectrometry (SELDI-TOF-MS) approach combined with bioinformatics. A total of 178 samples were analyzed in this study from untreated early stage DLBCL patients (38), patients with inflammatory lymphadenopathy (13), healthy donors (35), post-treatment non-relapsed DLBCL patients (53), and relapsed DLBCL patients (39). Model 1 formed by nine protein peaks (m/z: 6443, 5913, 6198, 4098, 7775, 9293, 5946, 5977, and 4628) could be used to distinguish DLBCL patients from healthy individuals with an accuracy of 95.89 % (70/73). The diagnostic pattern constructed using the support vector machine including the nine proteins of model 1, showed a maximum Youden's Index. Model 2 formed by three protein peaks (m/z: 3942, 6639, and 4121) could be used to distinguish DLBCL patients from those with inflammatory lymphadenopathy with an accuracy of 94.12 % (48/51). Model 3 formed by six protein peaks could distinguish patients with inflammatory lymphadenopathy from healthy individuals with an accuracy of 97.92 % (47/48). Model 4 could be used to distinguish non-relapsed DLBCL patients from relapsed DLBCL patients with an accuracy of 84.78 % (78/92). The four patterns were validated by leave-one-out cross-validation. These data demonstrate that the CM10 ProteinChip and SELDI-TOF-MS approach combined with bioinformatics can be used effectively to screen for the differential protein expression profiles of DLBCL patients and to predict the response to therapy.