ML-based sequential analysis to assist selection between VMP and RD for newly diagnosed multiple myeloma

Area of Science:

• Hematology
• Oncology
• Artificial Intelligence in Medicine

Background:

• Optimal first-line treatment is crucial for achieving deep and durable remission in newly diagnosed multiple myeloma (NDMM).
• Transplant-ineligible NDMM patients have limited treatment options, necessitating personalized therapeutic strategies.
• Predicting treatment response and survival outcomes is challenging in NDMM.

Purpose of the Study:

• To develop machine learning (ML) models for predicting overall survival (OS) and response in transplant-ineligible NDMM patients.
• To enable treatment-specific risk stratification for personalized therapy selection.
• To evaluate the potential impact of ML-guided treatment selection on patient outcomes.

Main Methods:

• Development of ML models using demographic and clinical characteristics at diagnosis.
• Training models to predict OS and response for two distinct first-line regimens: bortezomib plus melphalan plus prednisone (VMP) and lenalidomide plus dexamethasone (RD).
• Retrospective analysis to assess the impact of ML-guided treatment selection on survival and response.

Main Results:

• ML models enabled treatment-specific risk stratification, identifying patients likely to benefit from VMP or RD.
• Survival outcomes were superior when patients received the regimen for which they were classified as low risk.
• A significant survival benefit was observed in the VMP-low risk & RD-high risk group treated with VMP (Hazard Ratio: 0.15).
• Retrospective analysis suggested that ML models could have improved survival/response in 39% of patients (202 out of 514).

Conclusions:

• ML models trained on baseline clinical data can effectively stratify transplant-ineligible NDMM patients for first-line therapy.
• Individualized treatment selection guided by ML models has the potential to improve survival and response rates.
• This approach supports personalized medicine in NDMM, optimizing treatment strategies for better patient outcomes.