Screening of key genes associated with R-CHOP immunochemotherapy and construction of a prognostic risk model in diffuse large B-cell lymphoma

Diffuse large B-cell lymphoma (DLBCL) is a common subtype of non-Hodgkin lymphoma, which is curable in the majority of patients treated with rituximab plus cyclophosphamide, doxorubicin, vincristine and prednisone (R-CHOP) immunochemotherapy. However, the therapeutic mechanism of R-CHOP has not been elucidated. The GSE32918 and GSE57611 datasets were retrieved from The Gene Expression Omnibus database. The differentially expressed genes (DEGs) associated with R-CHOP therapy were identified using limma. Combined with prognostic information in GSE32918, DEGs found to be significantly associated with prognosis were selected using univariate Cox regression analysis and a risk prediction model was constructed. Based on this model, the samples in the training set (GSE32918) were divided into high and low risk score groups according to the median risk score. A total of 801 DEGs were identified between the R-CHOP treated DLBCL and primary DLBCL samples, from this 116 prognosis-associated genes were selected. Using Cox proportional hazards model, an optimal combination of 12 genes [including calcium/calmodulin dependent protein kinase I (CAMK1), hippocalcin like 4 (HPCAL4) and ephrin A5 (EFNA5)] was selected, and the sample risk score prediction model was constructed and validated. The DEGs between high risk score and low risk score groups were significantly enriched in functions associated with ‘response to DNA damage stimulus’, and pathways including ‘cytokine-cytokine receptor interaction’ and ‘cell cycle’. The optimal combination of the 12 genes, including CAMK1, HPCAL4 and EFNA5, was found to be useful in predicting the prognosis of patients with DLBCL after R-CHOP treatment. Therefore, these genes may be affected by R-CHOP in DLBCL.