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
Access, Representation, Population Health, Technology and Procedures, Imaging, Treatment Considerations, Combination Therapy, Therapy Sequence, Biological Therapies, Antibody Therapy, Bispecific Antibody Therapy, (CAR)-T Cell Therapies, Immunotherapy, Monoclonal Antibody Therapy, Adverse Events, Artificial Intelligence (AI), Machine Learning, Biological Processes, Clinical Practice (Health Services and Quality), Clinical Guidelines, Practice Models, Diseases, Immunology, Neutropenia, Infectious Diseases, Lymphoid Malignancies, Lymphomas, Myeloid Malignancies, Measurable Residual Disease, Non-Hodgkin Lymphoma, B Cell Lymphoma, Indolent Lymphoma, Aggressive Lymphoma, Research, Clinical Research, Clinical Trials, Epidemiology, Health Outcomes Research, Patient-Reported Outcomes, Real-World Evidence, Registries, Study Population, Human, Adult, Elderly, Young Adult
PAPER

5313 Supervised machine learning algorithms identify baseline features that predict infection in patients with follicular lymphoma receiving frontline epcoritamab

Siri Sagedahl, BA, Danielle Wallace, MD2, Robert Redd, MS3, Heather A. Walker, BS, MPH4, William Glenn Olson5, Rafaella Tringale6, Vanessa Kats, RN7, Ioulia Vogiatzi, PhD8, Emily Sumpena, BA9, Audrey Sigmund, MD10, Inhye E Ahn, MD11, Amy Bessnow, MD, MPH12, Jennifer R. Brown, MD, PhD13, Jennifer L. Crombie, MD14, Matthew S Davids, MD, MMSc15, David C Fisher, MD16, Cynthia K. Hahn, MD, PhD17, Eric Jacobsen, MD18, Caron Jacobson, MD, MMSc19, Austin I Kim, MD20, Ann LaCasce, MD, MMSc21, Oreofe Odejide, MD, MPH22, Erin Parry, MD, PhD23, David Qualls, MD24, Christine Ryan, MD25, Aswin Sekar, MD, PhD26, Sumana Devata, MD27, Carla Casulo, MD28, Kami Maddocks, MD29, Philippe Armand, MD30, Yazeed Sawalha, MD31, Svitlana Tyekucheva, PhD32, Reid W Merryman, MD33, Mark Murakami, MD34
1 Dana-Farber Cancer Institute, Boston, Massachusetts, United States
2 Wilmot Cancer Institute, University of Rochester Medical Center, Rochester, New York, United States
3 Dana-Farber Cancer Institute, Boston, Massachusetts, United States
4 Dana-Farber Cancer Institute, Boston, Massachusetts, United States
5 Dana-Farber Cancer Institute, Boston, Massachusetts, United States
6 Dana-Farber Cancer Institute, Boston, Massachusetts, United States
7 Dana-Farber Cancer Institute, Boston, Massachusetts, United States
8 Dana-Farber Cancer Institute, Boston, Massachusetts, United States
9 Dana-Farber Cancer Institute, Boston, Massachusetts, United States
10 The Ohio State University, Columbus, Ohio, United States
11 Dana-Farber Cancer Institute, Boston, Massachusetts, United States
12 Dana-Farber Cancer Institute, Boston, Massachusetts, United States
13 Dana-Farber Cancer Institute, Boston, Massachusetts, United States
14 Dana-Farber Cancer Institute, Boston, Massachusetts, United States
15 Dana-Farber Cancer Institute, Boston, Massachusetts, United States
16 Dana-Farber Cancer Institute, Boston, Massachusetts, United States
17 Dana-Farber Cancer Institute, Boston, Massachusetts, United States
18 Dana-Farber Cancer Institute, Boston, Massachusetts, United States
19 Dana-Farber Cancer Institute, Boston, Massachusetts, United States
20 Dana-Farber Cancer Institute, Boston, Massachusetts, United States
21 Dana-Farber Cancer Institute, Boston, Massachusetts, United States
22 Dana-Farber Cancer Institute, Boston, Massachusetts, United States
23 Dana-Farber Cancer Institute, Boston, Massachusetts, United States
24 Dana-Farber Cancer Institute, Boston, Massachusetts, United States
25 Dana-Farber Cancer Institute, Boston, Massachusetts, United States
26 Dana-Farber Cancer Institute, Boston, Massachusetts, United States
27 Medical College of Wisconsin, Milwaukee, Wisconsin, United States
28 Wilmot Cancer Institute, University of Rochester Medical Center, Rochester, New York, United States
29 The Ohio State University, Columbus, Ohio, United States
30 Dana-Farber Cancer Institute, Boston, Massachusetts, United States, Mayo Clinic, Rochester, Minnesota, United States
31 The Ohio State University, Columbus, Ohio, United States
32 Dana-Farber Cancer Institute, Boston, Massachusetts, United States
33 Dana-Farber Cancer Institute, Boston, Massachusetts, United States
34 Dana-Farber Cancer Institute, Boston, Massachusetts, United States

Background: Bispecific antibodies (BsAbs) have demonstrated strong activity in relapsed/refractory and treatment-naive (TN) follicular lymphoma (FL); however, grade ≥3 (G3+) infections have been reported in 10-30% of patients. (Linton et al. Lancet Hematology, 2024; Brody et al. ASH abstract, 2024; Falchi et al. Blood, 2025; Falchi et al. The Lancet, 2025). Defining patients’ infection risk could inform patient selection and risk mitigation strategies to maximize the clinical benefit of BsAbs. We previously reported initial results of a multicenter, phase 2 trial of epcoritamab and rituximab for TN, high-tumor burden FL (Merryman, ASH abstract, 2025; NCT05783609) and now evaluate infection risk among participants. 

Methods: We evaluated 51 participants enrolled between July 2023 and January 2026. These include patients accrued before (n=33) and after (n=18) protocol amendments to reduce dexamethasone prophylaxis intensity during epcoritamab step-up dosing (SUD) and the frequency of epcoritamab to every four weeks in cycles 4 through 9. Infections were graded per CTCAEv5.0. We compared patients who experienced a G2+ infection to patients who completed all 9 cycles of protocol treatment without any G2+ infections. Patients also had to have available baseline data for demographic characteristics, tumor burden (GELF criteria), and laboratory values. These baseline features as well as protocol-defined steroid dosing intensity were analyzed using a 100-bootstrap resampled Least Absolute Shrinkage and Selection Operator (LASSO) model with a tuned penalty parameter to predict the binary outcome of a G2+ infection during the nine cycles of study treatment. Due to limited cohort size, only the top two features, ranked by absolute magnitude of the standardized coefficients, were then evaluated using a bootstrap resampled logistic regression model to validate their predictive association with infection occurrence. To account for variation in clinical severity among CTCAEv5.0 G2 events (e.g., antiviral therapy for modestly symptomatic infections positive by highly sensitive molecular assays), this analysis was repeated using "clinically meaningful" infections, defined as a patient experiencing either G3+ or recurrent infections as the binary outcome. Applying this definition restricted the secondary cohort to 45 patients.  

Results: With a median follow-up of 9 months (range 4, 12), 28 patients (54.9%) within the primary cohort (n=51) experienced at least one G2+ infection (median duration: 7.0 days [2,35]) during treatment. The most common infections were upper respiratory (n=15) and COVID-19 (n=9). Eight patients (17.8%) in the secondary cohort (n=45) experienced clinically severe or recurrent meaningful infections, including three (6.7%) with G3+ events (median duration: 4.0 days [2,16]), two with recurrent (4.4%) and three (6.7%) with both severe and recurrent infection(s). The most common clinically severe or recurrent meaningful infections were COVID-19 (n=2) and upper respiratory infections (n=2). The LASSO model predicting G2+ infection, with its penalty parameter tuned using the one-standard-error rule, achieved a mean ROC-AUC of 0.78 (SD=0.01) across 100 bootstrap resamples, identifying stage IV versus stage II-III and patient age >64 versus 18-64 as the two most significant risk factors. The LASSO model predicting clinically severe or recurrent meaningful infections yielded a lower mean ROC-AUC of 0.68 (SD=0.02) but demonstrated a similar ranking of important features. Of note, reduced dexamethasone premedication intensity emerged as a protective factor in both models but was excluded from the logistic regression due to limited cohort size. When evaluating the logistic regression model with the top two predictors across the 100 bootstrap resamples, the model achieved a mean ROC-AUC of 0.78 (SD=0.01) for G2+ infections and 0.65 (SD=0.02) for clinically meaningful infections. 

Conclusions: The LASSO-based feature selection and logistic regression modeling approach yielded consistent variable selection, including baseline age and stage, as well as encouraging performance in predicting treatment-emergent G2+ and clinically meaningful infections. While these results remain exploratory and hypothesis-generating pending validation in larger cohorts, these findings provide a foundation for ongoing pursuit of a clinically tractable framework to stratify infection risk among patients with FL receiving BsAb-based therapy. 

Siri Sagedahl, BA: No relevant conflicts of interest to declare.