(PA-167) Measurable Residual Disease (MRD) Dynamics Reveals a Therapeutic Vulnerability State for Early Immunotherapy Interception in Multiple Myeloma (MM)

Understanding the underlying dynamics of MRD resistance could help in the design of tailored strategies empowered to eradicate MRD and improve survival of MM patients (pts).

Methods:

MRD dynamics were defined using NGF and Connector in 539 MM pts with ≥3 assessments in the GEM2012-2014 and GEM2017FIT trials. Validation was performed in a real-world cohort of 249 Greek pts.

            WES and RNA seq was performed to compare MRD with paired diagnostic and relapse tumors FACSorted from 107 MM pts. The bone marrow cell composition in matched diagnostic and MRD samples was analyzed using NGF.

             For experimental validation, MIcγ1^huCRBN immunocompetent mice, which carry humanized CRBN^I391V and progressively develop human-like MM, were treated with VRD (as in the GEM2012-2014 trial) until persistent MRD, which drove relapse upon treatment discontinuation. To evaluate the impact in OS, murine anti-BCMA CAR T cells were injected i.v. at MRD vs relapse.

Results:

A total of 3892 assessments from 788 pts were computed to identify three subgroups with different MRD dynamics: sustained undetectable (29.5%), positive/stable (46%) and positive/evolving levels (24.5%). The respective PFS rates at 6 years were 83%, 74% and 11%, and of OS were 89%, 91% and 28%. Positive/evolving MRD levels was the strongest risk-factor for PFS (HR: 9.1, p< .001) and OS (HR: 31.6, p< .001) in multivariate models including transplant-eligibility and the R-ISS. Identical results were observed in trial and real-world cohorts.

            MRD genomic divergency increased with high-dose chemotherapy when compared to induction. Further genomic complexity was observed in paired tumors sequenced at relapsed vs MRD vs diagnosis. WGS of mouse MM cells revealed increased number of genetic alterations after VRD therapy, including mutations in Trp53, DNA repair, CRBN, and CRBN-target genes. These results showcase MRD as potentially more vulnerable to therapy than relapsed tumors.     

            MRD transcriptional adaption also peaked after high-dose chemotherapy when compared to induction. Both human and mouse MRD cells were enriched in hallmarks linked to cell cycle, KRAS signaling, TNFα signaling and inflammatory response. In parallel, there was evolving distribution of immune cells in the tumor microenvironment at diagnosis vs MRD vs relapse, including the expansion of clonal and dysfunctional CD8 T cells. In line with the genomic findings, transcriptional and immune dynamics point to MRD as an optimal state for early intervention. Accordingly, deeper and prolonged responses were observed in MIcγ1^huCRBN mice following MRD interception using anti-BCMA CAR T cells vs identical treatment at relapse, which led to improved OS (median 12 vs 9.5 months; p< .001).

Conclusions:

This study presents MRD kinetics as the most relevant prognostic factor in MM, which could be intercepted with T-cell redirecting therapy to prevent continuous genomic evolution and immune modulation from MRD to relapse.