Early predictors of treatment-free remission in chronic myeloid leukemia.

6565 Background: Tyrosine kinase inhibitors (TKIs) have significantly improved outcomes for chronic myeloid leukemia (CML), allowing most patients to achieve near-normal life expectancy. This has shifted treatment goals toward achieving treatment-free remission (TFR), particularly important for younger patients. Current criteria for TFR rely on sustained deep molecular remission after years of therapy, but early predictors are lacking. We aimed to identify factors predictive of TFR eligibility. Methods: We screened 780 patients with newly diagnosed chronic phase CML treated at The University of Texas MD Anderson Cancer Center from January 2012 to December 2023. Among them, 412 patients (53%) met the NCCN criteria for TFR (sustained MR4.5 for 2 years following 3 years of therapy). Thirty patients (4%) were excluded due to insufficient treatment duration. The remaining 338 patients (43%) formed the control cohort. We compared these cohorts and evaluated predictive factors for TFR eligibility using univariate and multivariate logistic regression, including factors with P < 0.1 in the multivariate model. Results: Patients in the TFR cohort were older (median age 51 vs. 46 years, P = 0.007) and had a different distribution of Sokal risk categories (P = 0.037): 67% low risk, 27% intermediate, and 6% high risk, compared to 64%, 25%, and 11% in the control group. TKI usage also differed (P = 0.01), with 32% vs. 41% receiving imatinib, 43% vs. 33% dasatinib, 19% receiving nilotinib in both groups, and 6% vs. 8% receiving ponatinib. BCR::ABL transcript distribution was significantly different (P < 0.001). The e13a2 transcript was more common in the control group (51% vs. 33%), while the e14a2 transcript predominated in the TFR group (46% vs. 32%). Co-occurrence of both transcripts was similar (21% vs. 16%), but other variants were rare and only observed in the control group (1% vs. 0%). Resistance mutations in the ABL gene were exclusively detected in the control group (13%, 44 patients). Univariate analysis identified older age (OR: 1.02; P = 0.001), BCR::ABL halving-time <30 days (OR: 4; P < 0.001), and achieving molecular milestones—transcript levels <10% IS at 3 months (OR: 11.4; P < 0.001), <1% IS at 6 months (OR: 10.5; P < 0.001), and <0.1% IS (MMR) at 1 year (OR: 6.5; P < 0.001)—as predictive factors for TFR eligibility. The e14a2 transcript (OR: 2.2; P < 0.001), co-occurrence of both transcripts (OR: 1.9; P = 0.001), and treatment with newer-generation TKIs vs. imatinib (OR: 1.5; P = 0.008) were also significant predictors. Multivariate analysis confirmed that older age (OR: 1.2; P = 0.045), halving-time <30 days (OR: 2; P = 0.041), achieving MMR at 1 year (OR: 5.6; P < 0.001), and the e14a2 transcript (OR: 1.6; P = 0.049) were independent predictors. Conclusions: Early predictors of TFR eligibility include older age, halving-time <30 days, MMR at 1 year, and the e14a2 transcript. These factors could guide prospective trial designs as early surrogates for TFR.