Immunohistochemical Subtype and Parameters of International Prognostic Index in the New Prognostic Model of Diffuse Large B-Cell Lymphoma

Abstract

Aim. To develop an integrated prognostic model of diffuse large B-cell lymphoma (DLBCL) on the basis of immunohis-tochemical tumor subtype and parameters of International Prognostic Index (IPI). Materials & Methods. Out of 104 DLBCL patients in the data base 81 (77.9 %) met the eligibility criteria. Median age was 58 years (range 23-83). All patients were treated with R-СНОР. The creation of overall survival (OS) prognostic model for DLBCL patients was based on machine learning with classification and regression trees. OS was analyzed using Kaplan-Meier method. Survival curves were compared by means of log rank test and hazard ratio (HR). Any test was considered significant if two-sided level of p < 0.05 was reached. Results. Following the developed model three groups of patients were identified: the 1st group of low risk (the combination of low, intermediate-low, and intermediate-high risks according to IPI and GCB subtype); the 2nd group of intermediate risk (the combination of low, intermediate-low, and intermediate-high risks according to IPI and non-GCB subtype); the 3d group of high risk (irrespective of subtype). In the group of low risk (n = 26) 2-year OS during the monitoring period was 100 %. In the group of intermediate risk (n = 34) median OS was not reached, 2-year OS was 74 %, and expected 5-year OS was 68 %. In the group of high risk (n = 21) median OS was 25 months, 2-year OS was 46 %, and expected 5-year OS was 37 % (log rank p < 0.0001). HR calculated for the high-risk group compared with the low- and intermediate-risk groups was 5.1 (95% CI 2.112.1; p = 0.0003). Conclusion. A new integrated system of DLBCL prognosis is suggested which includes IPI risk parameters and immunohistochemical subtype based on Hans algorithm. This prognostic system can be used in clinical practice for DLBCL patient stratification and risk-adapted therapy.