Classifying the Risk of Relapse in Multiple Myeloma

Abstract

Multiple myeloma is one of the blood malignancies characterized by neoplastic proliferation of blood plasma cells. It accounts for 10% of the hematologic cancers. There do exist targeted drugs for multiple myeloma and there has been a significant improvement in outcomes of the disease. However, it has been observed that the patients do experience a relapse to the disease within first couple of years of initial diagnosis. In this paper, we aim to identify the factors that are strong predictors of a relapse within a period of 18 months. We applied decision tree algorithm for classification of profiles into two groups – high risk and low risk. High risk group being the one where the patient experienced a relapse of the disease within 18 months of initial diagnosis. Low risk group is the patients which experienced a progression free survival beyond 18 months after initial diagnosis. The data for the model was taken from published clinical trials, relevant features were taken from the published clinical trials. From these features we considered the most influential features and applied decision tree classifier for classification of the profiles. The factors that have highest impact in increasing the risk of relapse are serum albumin level, response to initial therapy, increase in serum monoclonal component, serum calcium, international staging level at the time of first diagnosis, plasma cell levels in bone marrow and freely circulating plasma cells. We applied Kaplan Meier survival analysis to predict the probability of relapse and median time to relapse in both risk groups. These results can be used to provide customized drug combinations to enable better outcomes for the patients.