Simplicity at the cost of predictive accuracy in diffuse large B-cell lymphoma: a critical assessment of the R-IPI, IPI, and NCCN-IPI

The international prognostic index (IPI) and similar models form the cornerstone of clinical assessment in newly diagnosed diffuse large B-cell lymphoma (DLBCL). While being simple and convenient to use, their inadequate use of the available clinical data is a major weakness. In this study, we compared performance of the International Prognostic Index (IPI) and its variations (R-IPI and NCCN-IPI) to a Cox proportional hazards (CPH) model using the same covariates in nondichotomized form. All models were tested in 4863 newly diagnosed DLBCL patients from population-based Nordic registers. The CPH model led to a substantial increase in predictive accuracy as compared to conventional prognostic scores when evaluated by the area under the curve and other relevant tests. Furthermore, the generation of patient-specific survival curves rather than assigning patients to one of few predefined risk groups is a relevant step toward personalized management and treatment. A test-version is available on lymphomapredictor.org.