Development and validation of a model to predict platelet response to romiplostim in patients with lower‐risk myelodysplastic syndromes

Summary Low endogenous erythropoietin levels and limited red blood cell transfusion history can predict response to erythropoiesis‐stimulating agents in anaemic patients with myelodysplastic syndromes ( MDS ). The relationship between endogenous thrombopoietin ( THPO ) levels and platelet response to romiplostim is unknown. Variables including baseline endogenous THPO levels, transfusion needs, and platelet response were analysed in a randomized trial of 250 thrombocytopenic, lower‐risk MDS patients (International Prognostic Scoring System low/intermediate‐1). A predictive scoring system was developed based on log–likelihood ratios and logistic coefficients. Patients with HI –P (haematological improvement – platelets) responses had lower mean baseline THPO levels ( P  =   0·0497) and were more likely to have <6 platelet units transfused in the past year ( P  =   0·0027), as did patients with platelet responses ≥50% of weeks on romiplostim ( P  =   0·001 and P  =   0·0037, respectively). A model for predicting response to romiplostim was developed and validated in a separate MDS cohort ( N  =   72). Patients in low‐, intermediate‐, and high‐response groups had response rates of 17·4%, 29·6%, and 50·7%, respectively, for HI ‐P, and 17·4%, 33·8%, and 65·2%, respectively, for ≥50% response. For thrombocytopenic patients with lower‐risk MDS , lower baseline THPO levels (<500 pg/ml) and limited platelet transfusion history predicted a greater likelihood of a subsequent platelet response to romiplostim.