AUTHORS
Xi Yang, MD, Xiaoying Chen, MS2, Adrienne N. Nedved, PharmD, MPA3, Yucai Wang, MD, PhD4, Natalie S Grover, MD5, Stephanie Franco, MD6, Tenley Schwarz7, Matthew Schwede, MD8, Marshall McKenna9, Joanna M Rhodes, MD, MSCE10, Yun Kyoung Tiger, MD, PhD11, Nayef Abdel-Razeq, MD12, Muhamad Alhaj Moustafa, MD, MS13, Catherine Reinert14, Ajay Major, MD, MBA15, Daniel Clancy Trotier, MD16, Priyanka A. Pophali, MD17, Kelsey Baron, MD18, Boyu Hu, MD19, Sally Jones, MD20, Aditi Saha, MD21, Nikita K Dave, MD22, Daniel Landsburg, MD23, Megan Sears-Smith24, Nilanjan Ghosh, MD25, Kiarash Salafian, MD26, Emily Ayers, MD27, Suchitra Sundaram, MD28, Joshua D. Brody, MD29, Alyssa Gibson, MD30, Justin Kline, MD31, Javier L. Munoz, MD, MBA32, Rodolfo Garza Morales, MD33, Cyrus Khan, MD34, Eiraj Khan, MD35, Brenda Sofia Castillo, MD36, Stephen Douglas Smith, MD37, Fadzai Chinyengetere, MDPhD38, Sandeep Dave, MD39, Alex Niu, MD40, Francisco Hernandez-Ilizaliturri, MD41, Paolo Fabrizio Caimi, MD42, Brian Hill, MD, PhD43, Taylor R. Brooks, MD44, Yasmin H Karimi, MD45
AFFILIATIONS
1 University of Michigan, Ann Arbor, Michigan, United States
2 Cleveland Clinic, Cleveland, Ohio, United States
3 Mayo Clinic, Rochester, Minnesota, United States
4 Mayo Clinic, Rochester, Minnesota, United States
5 University of North Carolina Lineberger Comprehensive Cancer Center, Chapel Hill, North Carolina, United States
6 University of North Carolina Lineberger Comprehensive Cancer Center, Chapel Hill, North Carolina, United States
7 Swedish Cancer Institute, Seattle, Washington, United States
8 Swedish Cancer Institute, Seattle, Washington, United States
9 Rutgers Cancer Institute of New Jersey, New Brunswick, New Jersey, United States
10 Rutgers Cancer Institute of New Jersey, New Brunswick, New Jersey, United States
11 Mayo Clinic, Rochester, Minnesota, United States
12 Mayo Clinic, Jacksonville, Florida, United States
13 Mayo Clinic, Jacksonville, Florida, United States
14 University of Colorado Cancer Center, Aurora, Colorado, United States
15 University of Colorado Cancer Center, Aurora, Colorado, United States
16 University of Wisconsin Carbone Cancer Center, Madison, Wisconsin, United States
17 University of Wisconsin Carbone Cancer Center, Madison, Wisconsin, United States
18 University of Utah Huntsman Cancer Institute, Salt Lake City, Utah, United States
19 University of Utah Huntsman Cancer Institute, Salt Lake City, Utah, United States
20 University of Alabama at Birmingham, Birmingham, Alabama, United States
21 University of Alabama at Birmingham, Birmingham, Alabama, United States
22 University of Pennsylvania School of Medicine, Philadelphia, Pennsylvania, United States
23 University of Pennsylvania School of Medicine, Philadelphia, Pennsylvania, United States
24 Levine Cancer Institute, Charlotte, North Carolina, United States
25 Levine Cancer Institute, Charlotte, North Carolina, United States
26 University of Virginia, Charlottesville, Virginia, United States
27 University of Virginia, Charlottesville, Virginia, United States
28 Icahn School of Medicine at Mount Sinai, New York, New York, United States
29 Icahn School of Medicine at Mount Sinai, New York, New York, United States
30 University of Chicago Comprehensive Cancer Center, Chicago, Illinois, United States
31 University of Chicago Comprehensive Cancer Center, Chicago, Illinois, United States
32 Mayo Clinic, Phoenix, Arizona, United States
33 Mayo Clinic, Phoenix, Arizona, United States
34 Allegheny Health Network, Pittsburgh, Pennsylvania, United States
35 Allegheny Health Network, Pittsburgh, Pennsylvania, United States
36 Fred Hutchinson Cancer Center, Seattle, Washington, United States
37 Fred Hutchinson Cancer Center, Seattle, Washington, United States
38 Duke University, Durham, North Carolina, United States
39 Duke University, Durham, North Carolina, United States
40 Roswell Park Comprehensive Cancer Center, Buffalo, New York, United States
41 Roswell Park Comprehensive Cancer Center, Buffalo, New York, United States
42 Cleveland Clinic Taussig Cancer Institute, Cleveland, Ohio, United States
43 Cleveland Clinic Taussig Cancer Institute, Cleveland, Ohio, United States
44 Cleveland Clinic Taussig Cancer Institute, Cleveland, Ohio, United States
45 University of Michigan, Ann Arbor, Michigan, United States
ABSTRACT
Introduction: Cytokine release syndrome (CRS) and immune effector cell-associated neurotoxicity syndrome (ICANS) are the main adverse events of clinical interest in patients treated with CD20xCD3 bispecific antibodies (BsAb). A better understanding of risk factors may aid in identification and management of low vs. high-risk patients and increase the utilization of these therapies in community practices. Previously published models using clinical trial populations have included parameters that are not routinely obtained in the real-world clinical settings. We aimed to develop predictive models for CRS and ICANS with readily available clinical parameters.
Methods: We obtained baseline characteristics and safety outcomes from a multicenter retrospective cohort including patients from 21 U.S. centers with relapsed or refractory (R/R) large B-cell lymphoma (LBCLs) receiving commercially available epcoritamab (epco) or glofitamab (glofit) monotherapy between 2023 and 2025. Elastic Net penalized logistic regression was used for predictor selection and model development. A point-based risk score was derived from the final model coefficients, and mean predicted risk was summarized for each risk score. Model performance was evaluated using the optimism-corrected AUC, Brier score, sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV).
Results: As of May 31, 2025, 331 patients with R/R LBCLs were treated with BsAb (epco n=203 [61%], glofit n= 128 [39%]); median age was 68; 64% male; 71% ECOG 0-1. Most common histologic type was diffuse large B-cell lymphoma, not otherwise specified (n=252, 83%) followed by high-grade B-cell lymphoma (n=53, 17%). Median number of prior lines of treatments was 3 (range 2- 4). Any grade CRS occurred in 120 (39%) patients: grade 1 (n=77, 25%), grade 2 (n=26, 8.5%), grade 3 (n=11, 3.6%), grade 4 (n=3, 1.0%), and grade 5 (n=3, 1.0%). ICANS occurred in 38 (13%) patients: grade 1 (n=16, 5.2%), grade 2 (n=9, 3.0%), grade 3 (n=7, 2.3%), grade 4 (n=3, 1.0%) and grade 5 (n=3, 1.0%).
For CRS, variables selected in the model included elevated LDH, bulky disease ³7.5cm, and advanced stage disease, resulting in an AUC of 0.57. Patients were assigned 1 point for abnormal LDH, and 2 points each for bulky or advanced stage disease for a total score between 0-5 points. Patients with a score of 0-1 were classified as low risk and had a mean predicted risk of CRS of 29%. Patients with a score of 2-5 were classified as high-risk and had a mean predicted risk of CRS of 39%. Using a score of 2–5 to define a positive test yielded a sensitivity of 92%, specificity of 20%, PPV of 41% and NPV of 81%. Internal validation resulted in Brier score of 0.23.
Given the clinical impact of high grade CRS on patient care, we next developed a model to predict ≥ Grade 2 (high grade) CRS. For high grade CRS, variables selected included bulky disease ≥ 7.5cm (1 point) and extra-nodal sites ≥2 (1 point), resulting in an AUC of 0.58. Using a score of 2 to define a positive test yielded a sensitivity of 23%, specificity of 89%, PPV of 23% and NPV of 89%. Internal validation resulted in Brier score of 0.1.
For ICANS, variables selected in the model and respective points assigned included: age ³60 (2 points), female gender (1 point), ECOG performance status >1 (1 point), bulky disease ³7.5cm (3 points), creatinine ≥ 1.5 x ULN (4 points), ALT or AST ≥ 3 x ULN (5 points) resulting in an AUC of 0.75. Points for each risk factor were additive and led to scores between 0-16. Patients were classified as low, intermediate or high risk for ICANS based on scores of 0-3, 4-8 or 9-16, with a mean predicted risk of ICANs of 6%, 21%, and 58%, respectively. Classifying intermediate- and high-risk patients as test positive yielded a sensitivity of 64.3%, specificity of 76.1%, PPV of 26.1%, and NPV of 94.2%. Restricting the positive group to high-risk patients increased specificity to 99.1% and PPV to 66.7% but reduced sensitivity to 14.3% and NPV to 89.8%. Internal validation resulted in Brier score of 0.08.
Conclusions: Use of clinically available parameters can identify patients at higher risk for CRS and ICANS. In low risk patients, this risk scoring model may allow clinicians to identify patients who can be safely treated with BsAb in the outpatient or community settings. External validation of this model in patients treated with both BsAb monotherapy and BsAB in combination with chemotherapy is ongoing.
DISCLOSURES
Xi Yang, MD: No relevant conflicts of interest to declare.