AI Multiple Myeloma Model Predicts Individual Risk, Outcomes, and Genomic Implications
A new machine-learning individualized prediction model identified 90 driver genes and 12 genomic subtypes for multiple myeloma and was also superior to current prognostic models, researchers reported (https://doi.org/10.1200/JCO.23.01277) in the Journal of Clinical Oncology. The study represents a major step forward in the treatment of multiple myeloma, which has highly variable presentation and outcomes.
Using genomic, demographic, clinical, and therapeutic data from 1,933 patients with newly diagnosed multiple myeloma, researchers trained the Individualized Risk in Newly Diagnosed Multiple Myeloma model (IRMMa) to predict a specific patient’s treatment outcomes and overall survival. To validate IRMMa, researchers used a separate data set (N = 256) of patients with newly diagnosed myeloma who were enrolled in the GMMG-HD6 trial (https://www.clinicaltrials.gov/study/NCT02495922).
In the modeling, researchers identified 90 driver genes whose variants were associated with myeloma, 10 of which were reported for the first time. Notably, 79% of participants had at least one nonsynonymous substitution (a variant that changes an amino acid in a protein) in at least one of the driver genes. Researchers also identified 12 distinct subtypes of myeloma, based on shared variants—another newly reported discovery.
“IRMMa's accuracy was significantly higher [in predicting overall survival and event-free survival] than all existing prognostic models,” researchers said (https://doi.org/10.1200/JCO.23.01277), such as the International Staging System and its two revisions.
However, compared to standard models, the researchers cautioned (https://doi.org/10.1200/JCO.23.01277) that “although recently proposed prognostic models . . . can identify a subgroup of high-risk patients, they are not corrected for different treatment approaches and are not designed to predict patient-level individual risk.”
This sets the stage for future application of precision medicine in multiple myeloma treatment. “The ability to capture each patient's specific treatment variance represents a critical tool that can help to select the most effective therapy and to avoid overtreatment where it adds little to no benefit,” the researchers concluded (https://doi.org/10.1200/JCO.23.01277).
Multiple myeloma is a complex and variable cancer. Learn more about the disease, treatments, and nursing considerations through the ONS resources listed in the sidebar.