A clinical prediction model for infusion-related reactions to rituximab in patients with B cell lymphomas

Background Infusion-related reactions (IRRs) are a major adverse event of rituximab. Objective To develop a prediction model for IRRs to rituximab among patients with B cell non- Hodgkin's lymphomas (B-NHL). Setting A 1000-bed university hospital in Tokyo. Methods Patients with B-NHL treated with rituximab at our institution from 2004 to 2014 were retrospectively analysed. Chills, fever, rash, nausea, asthenia, headache, cardiovascular symptoms, and respiratory symptoms of any grade, in association with rituximab infusion, were identified as IRRs. Risk factors for IRRs to rituximab found in the intergroup analysis were subsequently evaluated by using multivariate analysis. Main outcome measure Occurrence of IRRs to rituximab. Results A total of 140 patients with various types of B-NHL, including 74% with diffuse large Bcell lymphoma, were analysed. Among them, 55 and 85 patients were assigned to the IRR group and the non-IRR group, respectively. Indolent histological subtypes, bulky disease (>10 cm), B symptoms, higher serum soluble interleukin-2 receptor concentration, and bone marrow involvement were more common in the IRR group. The multivariate logistic regression analysis identified low-grade lymphomas [odds ratio (OR) 2.81, p = 0.017] and bulky disease (OR 2.52, p = 0.037) as independent risk factors for IRRs to rituximab. The incidence rates of IRRs to rituximab among patients with neither, one, or both of these risk factors were 26, 54, and 78%, respectively (χ² = 16.4, p < 0.001). Conclusions A simple combination of histopathological subtype and bulkiness of disease could predict the risk of IRRs to rituximab among patients with B-NHL.