Modelling MRD changes in myeloma to understand treatment effects, predict outcomes, and investigate curative potential.

Abstract

Purpose: We designed mathematical models to describe and quantify the mechanisms and dynamics of minimal residual disease (MRD) in order to better understand these MRD dynamics, to inform future treatment design, including when to stop or change treatment, and to extrapolate from current PFS times to predict future PFS curves.

Experimental design: To model individual sequential MRD data from phase III clinical trials (MAIA, CASTOR, and POLLUX) using previously developed mathematical models which would be modified as necessary to accurately correspond with the actual MRD data. These models would then be used to extrapolate PFS curves ahead in time.

Results: Patients with low MRD values either showed rapid disease regrowth or the MRD values remained low for a prolonged period. Treatment appeared to be most effective in terms of cell-kill within the first 6 to 12 months. Regrowth rates were correlated with estimated initial residual disease, particularly in MRD negative patients. Three-year model extrapolations of PFS were closely comparable to clinical trial data.

Conclusions: This model could provide early prediction of PFS outcomes, which otherwise takes lengthy periods of time to observe with clinical trials. Patients showing rapid rebound from low MRD values may benefit from adding another treatment before reaching progressive disease. The MRD analyses and results presented, such as the results about efficacy occurring early in the first 6 to 12 months, may help guide the development and selection of optimal regimens. Longer follow-up periods and application to other trials and datasets are required to substantiate these findings.