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, Pathology, Treatment Considerations, Combination Therapy, Supportive Care, Therapy Sequence, Biological Therapies, Transplantation, Checkpoint Inhibitor, Monoclonal Antibody Therapy, Non-Biological Therapies, Anticoagulant Drugs, Chemotherapy, Adverse Events, Artificial Intelligence (AI), Generative AI, Natural Language Processing, Clinical Practice (Health Services and Quality), Clinical Guidelines, Patient Safety, Quality Improvement, Diseases, Lymphoid Malignancies, Lymphoid Leukemias, Lymphomas, LGL, Non-Hodgkin Lymphoma, T Cell Lymphoma, Aggressive Lymphoma, Research, Clinical Research, Clinical Trials, Epidemiology, Health Outcomes Research, Real-World Evidence, Registries, Survivorship, Study Population, Human, Adult
PAPER

5358 Application of diagnostic criteria for T-large granular lymphocytic leukemia: A comparative analysis of contemporary classification schemes

Mahsa Hajizadeh, MD, Nicholas M. McCann, MD2, Benjamin Robert Highland, MD, MTheol3, Sahil Chaudhary, MBBS4, Bryna Shemo, MA5, David Feith, PhD6, Thomas P Loughran, MD7, Jeffrey W. Craig, MD, PhD8
1 Department of Pathology, University of Virginia Health System (UVA Health), Charlottesville, Virginia, United States
2 University of Virginia School of Medicine, Charlottesville, Virginia, United States
3 Department of Pathology, University of Virginia Health System (UVA Health), Charlottesville, Virginia, United States
4 Department of Pathology, University of Virginia Health System (UVA Health), Charlottesville, Virginia, United States
5 Department of Medicine, Division of Hematology/Oncology, University of Virginia School of Medicine, Charlottesville, Virginia, United States
6 Department of Medicine, Division of Hematology/Oncology, University of Virginia School of Medicine, Charlottesville, Virginia, United States, University of Virginia Comprehensive Cancer Center, Charlottesville, Virginia, United States
7 Department of Medicine, Division of Hematology/Oncology, University of Virginia School of Medicine, Charlottesville, Virginia, United States, University of Virginia Comprehensive Cancer Center, Charlottesville, Virginia, United States
8 Department of Pathology, University of Virginia Health System (UVA Health), Charlottesville, Virginia, United States, University of Virginia Comprehensive Cancer Center, Charlottesville, Virginia, United States

Distinguishing T-large granular lymphocytic leukemia (T-LGLL) from reactive large granular lymphocyte (LGL) expansions is essential for providing optimal patient care. To this end, 6 diagnostic schemes have been proposed for T-LGLL in the last half decade: WHO-HAEM5, ICC-2022, Shi/Morice, Jaffe, SH/EAHP, and the new international LGLL Consensus document. Here we provide the first practical demonstration of clinical efficacy for these schemes by performing head-to-head comparisons using data from the University of Virginia (UVA) LGLL Registry (encompassing patients with both established and suspected T-LGLL).

Of 647 patients meeting the essential National Comprehensive Cancer Network diagnostic testing requirements for T-LGLL, 384 (59%) had sufficient clinical and laboratory data to fully apply all 6 schemes (including standardized T-cell receptor gamma clonality assays, T-LGLL-focused flow cytometry, original bone marrow biopsy reports, and research-grade STAT3/STAT5B mutation sequencing). Cytopenia(s) were present in 339 patients (88%), and high-level STAT3/STAT5B mutations were found in 102 (27%). Two hundred eleven (55%) patients went on to receive first-line T-LGLL-directed therapy. Only 81 patients (21%) met the most restrictive LGL peripheral burden cutoff of >2,000 LGLs/µL, whereas 68% of patients had >500 LGLs/µL. T-LGLL was diagnosed in 111 (29%), 63 (16%), 46 (12%), 34 (9%), 30 (8%), 33 (9%) and 67 (17%) patients by 6, 5, 4, 3, 2, 1 and 0 schemes, respectively, equating to a moderate level of inter-scheme agreement (Fleiss κ=0.50).

As the primary measure of scheme performance, we compared how often each scheme rendered diagnoses in clinically relevant subsets of patients with suspected T-LGLL, including those with cytopenia(s), high-level STAT3/STAT5B mutations, or the subsequent need for first-line T-LGLL-directed therapy by calculating likelihood ratios (LR; for receiving a T-LGLL diagnosis in each relevant patient subset versus the remainder of the cohort) and odds ratios (OR; of belonging to each relevant subset among patients receiving a diagnosis versus not). All 6 schemes yielded positive LRs and ORs (with 95% confidence intervals exceeding 1.0) for patients with the high-level STAT3/STAT5B mutations and for those requiring first-line therapy. Although the new Consensus scheme had both highest LR (1.56; 95% confidence intervals (CI), 1.30–1.87) and OR (2.99; 95% CI, 1.96–4.58) for predicting first-line treatment need, no scheme was significantly superior in this regard. Focusing on patients with a positive response to one or more first-line therapies, 5 of 6 schemes (all except ICC-2022) yielded LRs and ORs >1.0, further supporting their clinical utility.

Based on the perception that absolute LGL thresholds may hinder recognition of T-LGLL with low peripheral blood burden, we hypothetically eliminated the requirement for 500 LGLs/µL imposed by some schemes, leading to new T-LGLL diagnoses (WHO-HAEM5, n=68; Shi/Morice, n=52; ICC-2022, n=50). These patients had similar frequencies of cytopenia(s), use of first-line therapy and positive treatment response compared to those diagnosed by the original schemes, but with a lower rate of STAT3/STAT5B mutations. Updated LRs and ORs were not statistically inferior to those of the original schemes, in keeping with preserved clinical performance.

By expanding our analysis to include the remaining 263 registry patients without complete data (and restricting our assessment solely to available clinical records), we identified T-LGLL-relevant mutational testing and bone marrow biopsy as the components of a comprehensive workup that are most often lacking (i.e., not performed) in patients narrowly missing out on a T-LGLL diagnosis. In terms of the percentage of nearly-diagnosed patients affected, missing mutational data most strongly impacted the Jaffe (85%), WHO-HAEM5 (84%), and Shi/Morice (83%) schemes, whereas the failure to collect a bone marrow biopsy was most problematic for Shi/Morice (45%), Jaffe (33%), ICC-2022 (31%) and SH/EAHP (29%).

In summary, this study provides the first comparative demonstration of clinical efficacy among proposed T-LGLL classification schemes and represents an evidence-based framework upon which future T-LGLL diagnostic criteria may be established. In particular, the recognition of clinically relevant “low-count” T-LGLL and the need to persue complete diagnostic workups when faced with potential T-LGLL emerge as immediate opportunities.

Mahsa Hajizadeh, MD: No relevant conflicts of interest to declare.