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

Paper Details

Abstracts, authors, affiliations, and disclosures.

SESSION

908. Outcomes Research: Myeloid Malignancies: Poster III

Program: Poster Abstract Session
Monday, December 14, 2026
06:00 p.m. - 08:00 p.m.
MCCNO - Hall F
Access, Representation, Population Health, Technology and Procedures, Molecular Testing, Emerging Technologies, Treatment Considerations, Combination Therapy, Drug Development, Supportive Care, Therapy Sequence, Biological Therapies, Transfusion, Vaccines, Non-Biological Therapies, Pharmacology, Adverse Events, Artificial Intelligence (AI), Natural Language Processing, Clinical Practice (Health Services and Quality), Clinical Guidelines, Health Economics, Patient Safety, Practice Models, Quality Improvement, Quality of Care, Systems-Based Hematology, Diseases, Lymphoid Malignancies, Lymphoid Leukemias, Myeloid Malignancies, Acute Myeloid Malignancies, Chronic Myeloid Malignancies, Measurable Residual Disease, ALL, Research, AML, APL, MDS, MPN, CML, Clinical Research, Clinical Trials, Epidemiology, Health Outcomes Research, Health Disparities Research, Patient-Reported Outcomes, Real-World Evidence, Registries, Survivorship, Study Population, Human, Adult, Elderly, Young Adult
PAPER

6327 AI-driven predictive model for treatment-free remission eligibility in patients with chronic myeloid leukemia

Talha Badar, MD, James Foran, FRCPC2, Cecilia Arana Yi, M.D.3, Gabriel Marquez4, Sunil Kumar Ravi5, Subash C6, Huichen Guan7, Randy Guerra8, Shashank Ghantoji, MD, PhD9, David Wei, PhD10, Cheng Shi11, Prerna Singh12, Praveen Kumar Marimuthu, MD13, Robert Matson14, Andrea Damon, PhD15, Kebede Begna, M.D.16
1 Mayo Clinic Comprehensive Cancer Centre, Jacksonville, Florida, United States
2 Mayo Clinic Comprehensive Cancer Centre, Jacksonville, Florida, United States
3 Mayo Clinic Comprehensive Cancer Centre, Phoenix, Arizona, United States
4 Novartis Ireland Ltd., Dublin, Ireland
5 Nference, Bengaluru, India
6 Nference, Bengaluru, India
7 Novartis Pharmaceuticals Corporation, East Hanover, New Jersey, United States
8 Novartis Pharmaceuticals Corporation, East Hanover, New Jersey, United States
9 Novartis Pharmaceuticals Corporation, East Hanover, New Jersey, United States
10 Novartis Pharmaceuticals Corporation, East Hanover, New Jersey, United States
11 Novartis Pharmaceuticals Corporation, East Hanover, New Jersey, United States
12 Nference, Bengaluru, India
13 Nference, Cambridge, Massachusetts, United States
14 Nference, Cambridge, Massachusetts, United States
15 Novartis Pharmaceuticals Corporation, East Hanover, New Jersey, United States
16 Mayo Clinic Comprehensive Cancer Centre, Rochester, Minnesota, United States

Introduction: In patients with chronic myeloid leukemia in chronic phase (CML-CP), treatment-free remission (TFR) is one of the primary long-term goals of therapy. TFR offers patients a chance to end reliance on daily medication and improve quality of life and reduce long-term treatment burden without concerning treatment-related adverse events. To be eligible for treatment-free remission (TFRe), patients must meet clinical readiness criteria: sustained deep molecular response (DMR) for at least 2 years and at least 3 years on TKI therapy (per NCCN 2026 CML guidelines). This analysis aims to identify the predictive factors that are associated with TFRe among patients with CML who achieved DMR when receiving care at the Mayo Clinic.

Methods: This retrospective cohort study examines CML-CP patients treated at the Mayo Clinic initiating first TKI (index) between May 2001 and December 2025. Patients were included if they (1) started an FDA approved index TKI, (2) were adults at index, (3) initiated index TKI within [-1, +6] months of initial diagnosis, and (4) had documented achievement of DMR (i.e., BCR::ABL(IS) ≤ 0.01%). De-identified patient data in nference's nSights platform were abstracted from structured EMR records and further extracted from unstructured EMR text (e.g. treatment interruptions, adverse events without diagnosis codes) using LLMs. Descriptive clinical and demographic statistics were summarized at baseline. Risk factors for time to TFRe were selected via LASSO regularization and analyzed with multivariate Cox regression. Risk factors included: age at diagnosis (18-39, 40-54, 55-64, 65-74, 75+), sex, race, diagnosis year (pre-2011, 2011-2016, post-2017), NCI comorbidity index (low: 0; intermediate: (0,0.55); high: ≥0.55), EMR (BCR::ABL1 (IS) ≤ 10% by 3 months), generation of TKI (1st, 2nd), major molecular response (MMR) by 12 months, time to DMR, adverse events (AEs), and dose modifications (interruption, dose reduction and escalation). Variable hazard ratios (HR) and 95% confidence intervals (CI) are reported.

Results: A total of 514 patients met inclusion criteria (median [IQR] age: 58.0 [48.0-68.0] years old, 40.1% female, 90.5% white, 94.6% insured, 50.6% high NCI, 39.9% 1st Generation TKI) with median [IQR] follow-up 87.0 [51.3-135.7] months. Patients were diagnosed at different time periods (pre-2011: 18.1%; 2011-2016: 32.9%; post-2017: 49.0%). Among patients who achieved DMR, 220 (42.7%) patients became TFRe.

Treatment modifications occurred in similar proportions of patients pre- and post-DMR achievement ([pre/post] dose reduction: 25.1%/33.3%; treatment interruption: 30.5%/28.4%; dose escalation: 14.6%/8.0%), with slightly more reductions post-DMR and slightly more escalations pre-DMR. The number of AEs reported post-DMR (median: 13) was comparable to pre-DMR (median: 12).

The multivariable Cox regression model estimated hazard ratios (HR) for the following five predictors of TFRe that were retained from the LASSO selection. (Model c-index after 5-fold cross validation: 0.680) Patients with CML diagnosis in recent years (post-2017 vs. pre-2011: HR [CI] 1.42 [1.16-1.74]) and patients having MMR at 12 months (HR [CI]:1.16 [0.97-1.39]) were more likely to achieve TFRe. However, patients diagnosed at younger ages (18-39 vs. 40-54: HR [CI] 0.82 [0.64-1.05]), patients who experienced dose escalation post-DMR (HR [CI]: 0.73 [0.54-1.00]), and patients with higher number of AEs (each additional AE: HR [CI] 0.98 [0.97-0.99]) were less likely to achieve TFRe.

Conclusions: After achieving DMR, patients with CML continued to report a high burden of tolerability-related events, including dose modifications and AEs, in this real-world analysis assisted with AI and LLM techniques. The predictive model further identified contemporary diagnosis and cumulative adverse-event burden—particularly after DMR—as independent predictors of TFR eligibility among patients who achieved DMR. Persistent treatment burden after DMR may influence progression to TFR eligibility and warrants consideration when long-term treatment goals include treatment discontinuation.

Talha Badar, MD: Syndax - Honoraria; Takeda - Honoraria; Ascentage - Honoraria; Geron - Honoraria; Amgen - Honoraria; Takeda - Research Funding; Amgen - Research Funding