(OA-03) Development and Validation of the SCOPE-MM Score: A Pre-Apheresis Risk Model Predicting Response and Toxicity in RRMM Patients Receiving BCMA-Directed CAR-T Therapy
Instructor Dana-Farber Cancer Institute Boston, Massachusetts
Introduction: Current risk stratification models for BCMA CAR-T in relapsed/refractory multiple myeloma (RRMM) primarily assess patients at lymphodepletion. However, this time point represents a ‘point of no return’ without the option of patient selection, bridging modification, or therapeutic planning. We aimed to overcome this limitation by identifying prognostic factors measurable at a pre-apheresis time point, allowing early prediction of CAR-T efficacy and toxicity.
Methods: In this multicenter observational study, demographic, laboratory, and clinical data were examined in a training cohort comprising 805 RRMM patients treated with either ciltacabtagene autoleucel (cilta-cel; n=291) or idecabtagene vicleucel (ide-cel; n=514) across 9 international sites. Multivariate Cox regression, time-dependent AUC analyses, and partitioning methods guided model development.
Results: Median age was 63 years (range 31–88), and patients had received a median of 5 (IQR 4-7) previous treatment lines. Multivariable analysis identified five pre-apheresis factors independently associated with inferior progression-free survival (PFS): elevated LDH levels, CAR-HEMATOTOX score ≥3, prior BCMA exposure, bone marrow plasma cell burden ≥50%, and history of extramedullary disease. Using these factors, we established the Stratification of CAR-T Outcomes at Pre-Apheresis Evaluation (SCOPE) score, assigning 2 points each for prior BCMA exposure and high CAR-HEMATOTOX score, and 1 point for each remaining factor.
Applying SCOPE, low-risk patients (0 points) demonstrated significantly superior outcomes compared to intermediate-risk (1-2 points) and high-risk (3-7 points) groups in terms of best objective response ([s]CR rate 58% vs 48% vs 35%, p< 0.001), 1-year PFS (71% vs 52% vs 27%, p< 0.001), and 1-year overall survival (93%, 80%, and 58%, p< 0.001). Furthermore, a distinct ultra-high-risk subgroup (SCOPE score ≥5), comprising 2.5% of patients, exhibited markedly poor outcomes (1y-PFS 5%). Subgroup analysis confirmed effective stratification across both products and geographic regions. Increasing SCOPE scores were associated with a significantly higher grade ≥3 rate forCRS (p=0.04), ICANS (p=0.03), early and late hematotoxicity (both p< 0.001), and infections (p=0.002). Higher scores also translated into increased non-relapse mortality (1y-NRM 1.7% vs 5.5% vs 9.6%, p< 0.001).
Importantly, the prognostic capacity of SCOPE was confirmed in two external validation cohorts, including a real-world dataset (n=135, PFS: p< 0.001) and the ide-cel arm of the KarMMa-3 trial (n=212, PFS: p< 0.001).
Conclusions: The comprehensively validated and product-agnostic SCOPE model effectively stratifies patients for toxicity and response with BCMA CAR-T before apheresis. This early assessment enables optimization of manufacturing logistics, informs outpatient management decisions, and helps identify high-risk patients who may benefit from more effective bridging, combinatorial strategies, or improved supportive care to mitigate NRM.
