PRIME Model
Predictive Relapse Indicators for Myeloma T-Cell Engagers (Blood 2025, ASH 2025)
Overview
PRIME (Predictive Relapse Indicators for Myeloma T-Cell Engagers) is a multivariable Cox model that predicts progression-free survival (PFS) in relapsed/refractory multiple myeloma patients receiving BCMA- and GPRC5D-targeting T-cell engagers. Published in Blood 2025 and presented at ASH 2025.
My Role: Contributing author on model development and validation.
The Problem
T-cell engagers (TCEs) targeting BCMA and GPRC5D have transformed RRMM treatment with unprecedented response rates. However:
- Many patients relapse early or show only transient responses
- No reliable predictors of treatment outcomes at TCE initiation
- Need for personalized risk assessment to tailor treatment strategies
Study Design
Retrospective cohort study at Mount Sinai Health System:
| Parameter | Value |
|---|---|
| Patients Identified | 325 |
| Patients Analyzed | 231 (minimal missing data) |
| PFS Events | 125 |
| Median PFS | 17.6 months (95% CI: 12.4-24.3) |
| Median Follow-up | 23.8 months (95% CI: 21.2-27.6) |
Cohort Characteristics
- Median age: 67.6 years
- High-risk cytogenetics: 42.4%
- Extramedullary disease (EMD): 22.1%
- Prior CAR-T: ~20%
- ≥5 prior lines of therapy: ~60%
- Triple-class refractory: 85.3%
- Penta-drug refractory: 39.8%
- BCMA-targeting TCE: ~55%
Statistical Methodology
Model Development
- Multivariable Cox model for PFS prediction
- 13 pre-specified clinical/laboratory predictors (21 degrees of freedom)
- Restricted cubic splines for continuous predictors (flexible modeling)
- Riley’s shrinkage methodology to mitigate overfitting
Validation
- Internal validation with optimism adjustment
- C-index: 0.659 (95% CI: 0.629-0.749)
Output
- Nomogram for individualized PFS estimates
- Predictions at 6, 12, 24, and 48 months
- Online calculator for clinical use (planned)
Key Predictors
Five predictors showed meaningful associations with PFS:
| Predictor | Comparison | Hazard Ratio | p-value | Interpretation |
|---|---|---|---|---|
| ALC | ≥0.9 vs <0.9 ×10³/µL | 0.62 (0.43-0.90) | 0.011 | 38% lower risk |
| High-risk cytogenetics | Present vs absent | 1.59 (1.10-2.30) | 0.015 | 59% higher risk |
| EMD | Present vs absent | 1.54 (1.02-2.33) | 0.041 | 54% increased hazard |
| LDH | 600 vs 200 U/L | 1.39 (1.06-1.81) | 0.015 | Continuous effect |
| Ferritin | 1000 vs 100 ng/mL | 1.54 (1.01-2.34) | 0.045 | 54% increased hazard |
Clinical Impact
PRIME enables personalized risk assessment for patients on TCE therapy:
- BCMA-targeting agents: Teclistamab, elranatamab
- GPRC5D-targeting agents: Talquetamab
- Risk stratification at treatment initiation
- Monitoring strategy optimization based on predicted risk
- External validation planned with cohorts from other centers
Technical Stack
Statistics: Multivariable Cox regression
Splines: Restricted cubic splines
Shrinkage: Riley's methodology
Validation: Internal (optimism-adjusted)
Output: Nomogram, Online calculator
Patients: N=231 (from 325 identified)
Publication
- Blood 2025 - “Development of Predictive Relapse Indicators for Myeloma T Cell Engagers (PRIME) Model” (Contributing Author)
- ASH 2025 - Conference presentation
Links
Collaborators
- Tarek Mouhieddine (Lead Author)
- Mount Sinai Myeloma Program