Dynamic prediction of Progression based on Minimal Residual Disease (MRD) in Patients with Newly Diagnosed Transplant Eligible Multiple Myeloma.
Flow cytometry and next-generation sequencing (NGS) have enabled sensitive monitoring of minimal residual disease (MRD) in multiple myeloma (MM). Recent studies have shown a strong association between landmark and sustained MRD negativity with improved progression-free survival (PFS). However, the predictive value of long-term MRD dynamics, including conversion from MRD negativity to MRD positivity, is not well explored. In the current study, we addressed whether and to what extent MRD conversion from MRD negativity to MRD positivity using NGS with a sensitivity of 10-5 would predict for clinical progression in newly diagnosed MM (NDMM).
This study used data of the CASSIOPEIA IFM 2015-01/HOVON131 trial, which enrolled 1085 transplant-eligible NDMM patients aged 18-65 years, who were randomized for induction and consolidation with or without daratumumab in combination with VTd (D-VTd, n=543; VTd, n=542), followed by a second randomization between daratumumab maintenance or observation. MRD by NGS was assessed at multiple timepoints after induction and consolidation, and yearly during maintenance and long-term follow up. Longitudinal MRD measurements (NGS 10−5) were analyzed using a joint model by combining a negative binomial mixed model for the longitudinal MRD data and a Cox proportional hazards model for time-to-progression, calculated from the first MRD negative measurement. International Staging System (ISS) score, cytogenetic risk, treatment with or without daratumumab, and time to first MRD negative sample were included as covariates in the model.
583 (54%) NDMM patients achieved MRD negativity (by NGS at 10-5, regardless of response) at any phase of the trial and were included in the current analysis. Patients had a median number of 4 (range 1-9) longitudinal MRD measurements. Median follow-up from first randomization was 79.7 ([IQR] 75.1-85.5) months. MRD increase measured with NGS was associated with an increased risk of progression (hazard ratio [HR] log-scale MRD 1.53; 95% CI [1.42-1.70], p<0.001). The joint model including MRD dynamics for prediction of clinical progression showed high discrimination (C-index 0.87). The model further enabled an individualized prediction, by dynamically updating the risk of progression after each MRD measurement. The median time from MRD conversion to clinical progression was 17.7 months (95% CI 13.4-24.0).
Dynamic assessment of MRD conversion from negative to positive measured with NGS (sensitivity of 10-5) is a strong predictor for clinical progression in MM patients who have achieved MRD negativity upon treatment. Integration of sequential MRD measurements with other predictive factors yielded a robust prediction model. These results show the potential of continued MRD monitoring for individual-patient risk prediction in clinical practice and MRD-guided decision making.