Predictive models for response to recombinant human erythropoietin (rhEPO) in cancer patients

Abstract

Response rates vary considerably among cancer patients treated with similar regimens of recom binant human erythropoietin (EPO); conse quently, identifying possible prognostic factors for response is beneficial. Prognostic factors identified in clinical trials include baseline EPO levels, baseline observed-to-predicted ratio of EPO levels, and rapid elevation of hemoglobin levels, reticulocyte count, and soluble trans ferrin receptor. The onset of the acute-phase inflammatory response, which is characterized by elevated C-reactive protein, neopterin, and increased ferritin levels, has also been associ ated with cancer-related anemia. Various models incorporating these prognostic factors have been proposed to predict the value of recombi nant human EPO therapy in anemic cancer pa tients. The clinical studies evaluating predictive algorithms are based on retrospective analyses. Therefore, prospective studies are needed. These predictive algorithms should be designed to use parameters that are widely available and easily interpreted.