The BCR repertoire comparison, lymphoma typing model and OS predicted model in 5 different pathological lymphomas: T-LBL/ALL, PTCL-NOS, B-MCL, B-FL, and DLBCL

8059 Background: Lymphatic system cancer is characterized by high heterogeneity in histology and clinical manifestations, B-cell antigen receptor (BCR) plays a vital role in anti-tumor immune responses. This study aimed to compare the BCR repertoire and identify some specific immune markers for different pathological lymphomas. Methods: 5 pathological types of non-Hodgkin's lymphoma (T-LBL/ALL, PTCL-NOS, B-MCL, B-FL, DLBCL) were collected, with reactive lymph node (RLN) hyperplasia as control. All patients were tested by high-throughput immunohistochemical sequencing (HTS-IR) to analyze the correlation between B-cell immunohistochemistry and clinical indicators, and constructed new strategy typing and overall survival (OS) predicted models for lymphomas. Results: The BCR repertoire had the highest diversity in RLN, followed by T-LBL/ALL, PTCL-NOS, DLBCL, B-MCL and B-FL. The diversity of BCR repertoire and similarity of B cell antigens were higher in B-MCL and B-FL patients. Similar to RLN, T-LBL/ALL and PTCL-NOS had broad and diverse V-J pairs, and rare in B-MCL, B-FL and DLBCL. RLN patients were with the highest average number of amino acids, followed by T-LBL/ALL, DLBCL, PTCL-NOS, B-MCL and B-FL. The expressed amino acid sequencing of ARDLIALDY, ARRPGSFDY, ARDIAGWGAVAGLLGRAYYGMDV, and ARDGPYGGNSVEYFQH were markedly different among 5 groups. Patients tended to recurrence expressed ASLDSSPSGFC, ARGMTTVTTAPNY, ARVPLYDDQNINDV and AGGVGGYDWGSYYFDY (P = 0.01605, 0.02869, and 0.01569), and prone to metastasis with expressions of ARVKEFYGILTGYDY, AHSIIGSSWYNWFDP and VRDGGWQSNNWLGFDV (P = 0.04259, 0.0450 and 0.0481). For all patients, 18 (7 negative, 11 positive) and 12 (10 negative, 2 positive) IGH V-J pairs were respectively associated with lymphoma recurrence and metastasis. The top 3 most significant pairs were IGHV7-4-1_IGHJ4, IGHV3-53_IGHJ5 and IGHV3-7_IGHJ5 bound up with recurrence (P = 0.0019, 0.0020 and 0.0021), and IGHV3-74_IGHJ1, IGHV1-69_IGHJ3 and IGHV1-2_IGHJ1 related to metastasis (P = 0.0022, 0.010 and 0.019). The accuracy of typing model in training and test sets was 78.125% (25/32) and 60% (6/10), respectively. The OS model can predict long (≥ 24 months) or short ( < 24 months) OS. Conclusions: Our study identified new biomarkers, constructed novel lymphoma typing model and OS predicted model based on B cell repertoire. It provides a comprehensive understanding of immune response, and contributes to the diagnosis and prognosis of non-Hodgkin's lymphoma.