Dose reduction and toxicity of lenalidomide-dexamethasone in multiple myeloma: A machine-learning prediction model

Abstract

Purpose Lenalidomide remains an effective drug for multiple myeloma, but it is often associated with adverse events and requires dose adjustments. The objective of this study was to propose a model for predicting whether a patient would require dose adjustment. Methods This retrospective observational study included patients treated with lenalidomide and dexamethasone from June 2014 to September 2018 at a tertiary hospital. Demographic variables, patient functional status, disease, analytical data specific to myeloma, and treatment-related variables were collected. Univariate and machine learning (logistic regression and classification and regression trees model) analyses were also performed. Kaplan–Meier analysis was used to determine the time of toxicity onset. Only lenalidomide (and not dexamethasone) related dose reductions are included. Results A total of 64 patients received lenalidomide-dexamethasone. 69% (44) required dose reduction or discontinuation of treatment due to lenalidomide-related adverse events. The median time between treatment beginning and lenalidomide dose reduction or discontinuation was 8.0 months (95% CI: 6.0–17.0). Age, platelet count, and neutrophil count were related to dose reduction in the univariate model. In the multivariate models, age and neutrophil count were significant in the logistic regression model, renal clearance, and neutrophil count in the classification and regression trees model. Conclusion Elderly patients and those with low bone marrow reserves are prone to dose-limiting adverse events. This study can aid in making follow-up, prophylaxis, and dosing decisions to achieve better pharmacotherapeutic results.