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, Newborn Screening, Omics Technologies, Pathology, Serologic Tests, Bioinformatics, Computational Biology, Emerging Technologies, Profiling, Treatment Considerations, Combination Therapy, Therapy Sequence, Biological Therapies, Antibody Therapy, Transplantation, Bispecific Antibody Therapy, (CAR)-T Cell Therapies, Monoclonal Antibody Therapy, Non-Biological Therapies, Chemotherapy, Surgical, Adverse Events, Artificial Intelligence (AI), Computer Vision, Deep Learning, Generative AI, Machine Learning, Natural Language Processing, Biological Processes, Pathogenesis, CHIP, Genomics, Hematopoiesis, Immune Mechanism, Metabolism, Molecular Biology, Multi-Systemic Interactions, Clinical Practice (Health Services and Quality), Clinical Guidelines, Diseases, Hemoglobinopathies, Sickle Cell Disease, Immunology, GVHD, Lymphoid Malignancies, Lymphoid Leukemias, Lymphomas, Plasma Cell Disorders, Myeloid Malignancies, Acute Myeloid Malignancies, Chronic Myeloid Malignancies, Measurable Residual Disease, ALL, CLL, Non-Hodgkin Lymphoma, B Cell Lymphoma, Indolent Lymphoma, Aggressive Lymphoma, Research, Fundamental Science, AML, APL, Translational Research, MDS, MPN, CMML, Clinical Research, Clinical Trials, Health Outcomes Research, Patient-Reported Outcomes, Real-World Evidence, Registries, Survivorship, Study Population, Human, Adult, Elderly, Pediatric
PAPER

6120 Pathology-correlated PET radiomics to differentiate follicular lymphoma progression from histologic transformation at the single lesion level

Efrat Luttwak, Luttwak, MD, Jonathan Kuten2, Annie Qiu, MPH3, Charlie White4, Parnian Shobeiri5, Michelle Okwali, MPH6, Kevin Han7, Daniel Lafontaine8, Suheil Albert Atallah-Yunes, MD9, Kevin David, MD10, Zachary Epstein-Peterson, MD11, Lorenzo Falchi, MD12, Paola Ghione13, Steven Horwitz, M.D.14, Anita Kumar, MD15, Jennifer Kimberly Lue, M.D.16, Alison Moskowitz, MD17, Ariela Noy, MD18, Colette Owens, MD19, Maria Lia Palomba, MD20, Mark Roschewski, MD21, Quinlan Sievers, MD, PhD22, Raphael Steiner, MD23, Robert Stuver, MD24, Pallawi Torka, MD25, Jasmine Zain, M.D.26, Andrew Zelenetz, MD27, Audrey Mauguen28, Heiko Schoder, MD29, Gilles Salles I, MD, PhD30
1 Memorial Sloan Kettering, NYC, New York, United States
2 Memorial Sloan Kettering, NYC, United States
3 Memorial Sloan Kettering, NYC, New York, United States
4 Memorial Sloan Kettering, NYC, United States
5 Memorial Sloan Kettering, NYC, United States
6 Memorial Sloan Kettering, NYC, New York, United States
7 University of California San Diego, La Jolla, California, United States
8 Memorial Sloan Kettering, NYC, United States
9 Memorial Sloan Kettering, NYC, New York, United States
10 Memorial Sloan Kettering, NYC, New York, United States
11 Memorial Sloan Kettering, NYC, New York, United States
12 Memorial Sloan Kettering, NYC, New York, United States
13 Memorial Sloan Kettering, NYC, New York, United States
14 Memorial Sloan Kettering, NYC, New York, United States
15 Memorial Sloan Kettering, NYC, New York, United States
16 Memorial Sloan Kettering, NYC, New York, United States
17 Memorial Sloan Kettering, NYC, New York, United States
18 Memorial Sloan Kettering, NYC, New York, United States
19 Memorial Sloan Kettering, NYC, New York, United States
20 Memorial Sloan Kettering, NYC, New York, United States
21 Memorial Sloan Kettering, NYC, New York, United States
22 Memorial Sloan Kettering, NYC, New York, United States
23 Memorial Sloan Kettering, NYC, New York, United States
24 Memorial Sloan Kettering, NYC, New York, United States
25 Memorial Sloan Kettering, NYC, New York, United States
26 Memorial Sloan Kettering, NYC, New York, United States
27 Memorial Sloan Kettering, NYC, New York, United States
28 Memorial Sloan Kettering, NYC, United States
29 Memorial Sloan Kettering, NYC, United States
30 Memorial Sloan Kettering, NYC, New York, United States

Introduction

Follicular lymphoma (FL) is an indolent B-cell lymphoma with patients experiencing generally favorable outcomes; however, some patients undergo histologic transformation (tFL) to aggressive lymphoma, significantly impacting patients’ prognosis. Distinguishing non-transformed progression (pFL) from transformation (tFL) remains clinically challenging, and while high SUV on FDG-PET has been proposed to raise suspicion for transformation, its predictive value has varied across studies. Unlike prior approaches based on whole-body maximum SUV, we assessed radiomic features at the single-lesion level, spatially matched to the pathologically confirmed biopsy site, enabling direct radiologic–pathologic correlation. We evaluated whether lesion-level FDG-PET radiomic features discriminate biopsy-proven pFL versus tFL.

 Methods

This retrospective, single-institution study included patients with FL grade 1-3A on active surveillance after initial diagnosis who subsequently developed biopsy-proven progression or transformation. Exclusion criteria included FL grade 3B, transformation occurring less than 3 months after diagnosis, no PET/CT at the time of pFL/tFL, and absence of FDG-avid targetable lesions for segmentation. For each patient, a single lesion corresponding to the subsequently pathologically confirmed site was segmented and radiomic features (first-order [intensity/histogram], shape, and texture) were extracted. In all patients, the lesion selected for biopsy was the most metabolically active (highest-SUV) safely targetable lesion, sampled for clinically suspected progression regardless of eventual pathology. Clinical variables at diagnosis and at progression were also assessed. Univariable logistic regression assessed each standardized radiomic feature's association with transformation, with Benjamini-Hochberg correction to control for false discovery rate. Classification trees, which recursively partition patients using feature thresholds to assign each lesion a predicted class (pFL or tFL), were built from radiomic features alone and from combined radiomic and clinical features.

Results

We identified 169 patients (median age 59 years [IQR 52-68]; 83% Lugano stage III/IV at diagnosis; median 2.2 years [IQR 1.0-4.3] from diagnosis to the index event, pFL or tFL); of them, 119 had pFL and 50 had tFL (30%). Numerous lesion-level radiomic features across first-order, shape, and texture classes were significantly associated with transformation (q<0.05).

SUVmax of the segmented lesion, a first-order radiomic measurement, discriminated transformation with an AUC of 0.86 (95% CI 0.79-0.93); a SUVmax threshold above 15 yielded 90% sensitivity and 75% specificity.

Texture features quantifying intratumoral heterogeneity showed the strongest associations, led by gray-level size-zone nonuniformity (OR 11.5) and high-gray-level zone emphasis (OR 9.66). First-order features, including energy (OR 6.80) and range (OR 6.16), and shape features, including least axis length (OR 1.70), were also associated with transformation, while higher uniformity conferred lower odds (OR 0.14).

A classification tree based on radiomic features alone correctly classified 93% of pFL cases (111/119) and 80% of tFL cases (40/50), corresponding to a positive predictive value (PPV) of 83% and a negative predictive value (NPV) of 92% for transformation.

Among clinical variables, older age (OR 1.04), FL grade 3A at diagnosis (OR 3.07), elevated LDH at the index event (OR 3.57), and an extranodal segmented site (extranodal vs nodal, OR 4.76) were associated with transformation.

Adding clinical variables improved classification, with a combined radiomic–clinical tree correctly classifying 95% of pFL cases (113/119) and 90% of tFL cases (45/50), corresponding to a PPV of 88% and an NPV of 96% for transformation.

Conclusions

Lesion-level FDG-PET radiomic features discriminate pFL from tFL with high accuracy, with greater intratumoral metabolic heterogeneity, in addition to SUVmax, associated with transformation. A radiomic signature may serve as a non-invasive tool to complement clinical assessment. By correctly classifying the majority of progressions and transformations, with NPV of 96% and PPV of 88% in this cohort, which had a high rate of tFL, the signature may help raise or lower suspicion for transformation and guide targeted biopsy decisions.

Efrat Luttwak, Luttwak, MD: ADC therapeutics - Consultancy (Includes expert testimony)