Multiomic profiles predict MRD status after quadruplet therapy in myeloma

Researchers performed genomic, transcriptomic and DNA methylation profiling of 423 tumor samples from the MIDAS cohort. All patients received induction with isatuximab, carfilzomib, lenalidomide and dexamethasone (Isa-KRd); molecular predictors of MRD were assessed among trial-enrolled patients. The analysis included whole-genome sequencing, RNA sequencing and methylation sequencing, with all three data types available for 416 samples. Molecular subgroups and features were evaluated in external myeloma cohorts.

Postinduction MRD-negativity rates differed across gene-expression subtypes: HP1, HP3 and MS groups had better responses than MF and CD2. B-cell differentiation signatures were associated with MRD positivity, whereas proliferation and MYC programs were associated with negativity. A multimodal random-forest model integrating all three molecular layers predicted MRD status with an AUC of 0.77, outperforming clinical and genomic models; its discrimination was comparable to transcriptomic and epigenomic models, with improved calibration.

The findings describe molecular variation associated with response to Isa-KRd and may inform future risk-stratification research. The model is not presented as a validated clinical decision tool, and MRD is an early response endpoint; the study does not establish that changing treatment based on these profiles improves patient outcomes.