A Threshold Regression Mixture Model for Assessing Treatment Efficacy in a Multiple Myeloma Clinical Trial