2026 ASH®
Annual Meeting & Exposition December 12-15, 2026 ▪ New Orleans, Louisiana

Paper Details

Abstracts, authors, affiliations, and disclosures.

Monday, December 14, 2026
06:00 p.m. - 08:00 p.m.
MCCNO - Hall F
Technology and Procedures, Assays, Imaging, Molecular Testing, Omics Technologies, Bioinformatics, Computational Biology, Profiling, Treatment Considerations, Combination Therapy, Biological Therapies, Transplantation, Non-Biological Therapies, Chemotherapy, Artificial Intelligence (AI), Machine Learning, Biological Processes, Apoptosis, Pathogenesis, Genomics, Hematopoiesis, Immune Mechanism, Metabolism, Microbiome, Molecular Biology, Multi-Systemic Interactions, Clinical Practice (Health Services and Quality), Clinical Guidelines, Diseases, Bone Marrow Failure Syndromes, Acquired Marrow Failure Syndromes , Myeloid Malignancies, Acute Myeloid Malignancies, Chronic Myeloid Malignancies, Measurable Residual Disease, Research, Fundamental Science, AML, APL, Translational Research, MDS, Clinical Research, Real-World Evidence, Registries, Study Population, Human, Adult, Elderly, Pediatric, Young Adult, Animal Model
PAPER

5233 Applying the PRISM score to relapsed/refractory AML treated with HMA + venetoclax: A multi-center academic and nationwide real-world cohort analysis

Henry Dumke, MD, Parisa Abedi, MD2, Luca Lanino, MD3, Jan Philipp Bewersdorf, MD4, Tariq Kewan, MD5, Mrinal M. Patnaik, MD, MBBS6, Wei Wei, PhD7, tulika rudragupta, MS8, Rohan K Achar, MD, MSCR9, Nurefsan Sariipek, MD10, Razan Mansour, MD11, Benjamin Rolles, MD12, Abdulrahman Alhajahjeh, MD13, Alain Mina, MD14, Luis E. Aguirre, MD, MPH15, Nikolai Podoltsev, MD, PhD16, Lourdes Mendez, MD, Ph D17, Mina L Xu, MD18, Alexa Siddon, MD19, Alessandro Baldi, PhD20, Claudia Haferlach, MD21, Aref Al-Kali, M.D.22, Torsten Haferlach, MD, PhD23, Amer M. Zeidan, MD24, Maximilian Stahl, MD25
1 Yale School of Medicine, New Haven, Connecticut, United States
2 Yale School of Medicine, New Haven, Connecticut, United States
3 Yale School of Medicine, New Haven, Connecticut, United States
4 Yale School of Medicine, New Haven, Connecticut, United States
5 Mayo Clinic, Rochester, Minnesota, United States
6 Mayo Clinic, Rochester, Minnesota, United States
7 Yale School of Medicine, New Haven, Connecticut, United States
8 Yale School of Medicine, New Haven, Connecticut, United States
9 Yale School of Medicine, New Haven, Connecticut, United States
10 Yale School of Medicine, New Haven, Connecticut, United States
11 Yale School of Medicine, New Haven, Connecticut, United States
12 Yale School of Medicine, New Haven, Connecticut, United States
13 Yale School of Medicine, New Haven, Connecticut, United States
14 Yale School of Medicine, New Haven, Connecticut, United States
15 Yale School of Medicine, New Haven, Connecticut, United States
16 Yale School of Medicine, New Haven, Connecticut, United States
17 Yale School of Medicine, New Haven, Connecticut, United States
18 Yale School of Medicine, New Haven, Connecticut, United States
19 Yale School of Medicine, New Haven, Connecticut, United States
20 MLL Munich Leukemia Laboratory, Munich, Germany
21 MLL Munich Leukemia Laboratory, Munich, Germany
22 Mayo Clinic, Rochester, Minnesota, United States
23 MLL Munich Leukemia Laboratory, Munich, Germany
24 Yale School of Medicine, New Haven, Connecticut, United States
25 Yale School of Medicine, New Haven, Connecticut, United States

Introduction: Risk stratification for patients with acute myeloid leukemia (AML) treated with lower-intensity hypomethylating agents plus venetoclax (HMA+VEN) is an active area of research. Recently, the Prognostic Risk Integration for Survival Modeling (PRISM) score was developed to expand upon the 4-gene molecular classifier (FLT3-ITD, N/KRAS, TP53), improving overall survival (OS) prediction for newly diagnosed (ND)-AML (Lachowiez et al., JCO 2026). However, the prognostic utility of the PRISM score in relapsed/refractory (R/R)-AML patients treated with HMA+VEN remains unclear. To address this, we evaluated the validity of the PRISM score in a large cohort of 370 R/R-AML patients and compared its performance against a cohort of 2,283 ND-AML patients across academic and community settings. 

Methods: We analyzed patients with AML treated with HMA+VEN across two data sources: (1) the VALIDATE-AML consortium consisting of three academic centers (Yale Cancer Center, Mayo Clinic, and the Munich Leukemia Laboratory [MLL]) and (2) the nationwide real-world Flatiron Health Research database. The continuous PRISM-3 ridge linear predictor was applied unchanged to assign patients into low-, moderate-, and high-risk tertiles. Within each cohort, we stratified by clinical setting (ND vs. R/R). Prognostic discrimination of the continuous PRISM score for OS was quantified using Harrell's C-index estimated with bootstrap 95% confidence intervals (CIs); the ND-vs-R/R difference (ΔC) was estimated by bootstrap. Tier separation of survival was tested by log-rank.

Results: The VALIDATE-AML cohort included 491 ND and 191 R/R patients; the Flatiron cohort included 1,792 ND and 179 R/R patients. R/R patients had lower median PRISM scores than ND (VALIDATE-AML: 0.99 vs. 1.31, p<0.001; Flatiron: 1.24 vs. 1.39, p<0.001), corresponding to a shift in tier composition (VALIDATE-AML ND vs. R/R low/moderate/high: 31/36/33% vs. 65/26/9%; Flatiron ND vs. R/R low/moderate/high: 24/43/33% vs. 40/39/21%). R/R populations differed from ND on several PRISM inputs: R/R patients were younger than ND (median age 64 vs. 73 in VALIDATE-AML, p<0.001; 69 vs. 76 in Flatiron, p<0.001), with lower TP53 mutation prevalence (14% vs. 27% in VALIDATE-AML, p<0.001; 11% vs. 22% in Flatiron, p<0.001), and lower adverse karyotype in VALIDATE-AML (21% vs. 43%, p<0.001), but similar in Flatiron (31% vs. 34%, p=0.51).

Median OS declined monotonically across PRISM tiers in both settings. In VALIDATE-AML ND, median OS was 30.4, 13.5, and 7.3 months for low, moderate, and high PRISM risk tiers, respectively (log-rank p<0.001); in VALIDATE-AML R/R, median OS was 15.6 months for low (n=125), 12.2 months for moderate (n=49), and 3.5 months for high (n=17) risk tiers (log-rank p<0.001). In Flatiron ND, median OS was 18.8, 11.2, and 7.8 months (log-rank p<0.001); in Flatiron R/R, median OS was 12.3 months for low (n=72), 10.0 months for moderate (n=69), and 8.4 months for high (n=38), with tier separation reaching only borderline significance (log-rank p=0.052).

PRISM discrimination was higher in ND patients (VALIDATE-AML: C=0.64, 95% CI 0.62–0.67; Flatiron: C=0.61, 0.60–0.62), while it dropped substantially in the R/R setting (VALIDATE-AML: C=0.58, 0.54–0.62; Flatiron: C=0.55, 0.52–0.59), corresponding to a decrement of ΔC=0.064 (0.000–0.126) in VALIDATE-AML and ΔC=0.058 (0.005–0.112) in Flatiron.

Conclusions: We successfully validated the PRISM score in patients with ND-AML treated with HMA+VEN. However, its prognostic performance was weaker in the R/R-AML setting, indicating a shift in the drivers of outcome away from the features validated in ND-AML. These findings support either redefining the PRISM score cut points or developing a new risk model for R/R-AML treated with HMA+VEN.
Henry Dumke, MD: No relevant conflicts of interest to declare.