Background and purpose:

Treatment-free remission (TFR) has become a goal for chronic myeloid leukemia (CML) patients treated with tyrosine kinase inhibitors (TKIs) and maintain a deep molecular response (DMR). Some prognostic factors have been reported to influence and predict TFR outcomes. We aimed to investigate the real-world outcomes and to identify the predictors of successful TFR in patients who discontinued TKI therapy. Methods: Between 2002 and 2024, all newly diagnosed CML patients were treated with frontline TKIs and underwent BCR::ABL1 monitoring at the reference laboratory at Chang Gung Memorial Hospital-Linkou and followed at 6 medical centers in Taiwan. Totally, 101 adult CML patients stopping TKIs and fulfilling the criteria according to the guidelines of Hematology Society of Taiwan (adapted from the ELN recommendations 2020) were enrolled. All patients achieved BCR::ABL1IS > MR4.5 or deeper for at least 2 years; with imatinib treatment duration for >5 years or nilotinib/dasatinib for >4 years. Relapse was defined as loss of major molecular response (MMR). Patient-specific halving time for the first 3 months was available and calculated for 92 of 101 patients. Gene mutations were analyzed using SOPHiA Genetics Myeloid Solution v2 panel or RNA-based gene panel assay (Branford et al. J Mol Diagn 2022) on ASXL1 for whole cohort at diagnosisand NGS panel forrelapse or prior TKI failure patients at diagnosis, TFR initiation, and relapse. Results: At time ofTKI stop, the median age was 58.5 years, 49.4% were male, 65.3% had e14a2 transcripts. ELTS risks were low in 70.3%, intermediate 24.2%, high 5.5%. Before stopping TKIs, 53.5% were treated with imatinib, 18.8% with dasatinib, and 27.7% with nilotinib. Among the 81 patients with cytogenetic data, 79 were low-risk and 2 were intermediate-risk; none had high-risk features (ELN 2020). The median TKIs treatment duration was 136.6 months (imatinib 160.9 months, dasatinib 109.5 months, and nilotinib 97.3 months). All patients had sustained MR4.5 before TKI discontinuation (median 95.1 months) and 99 of them had undetectable BCR::ABL1 (median 78.5 months), and 76 reached MR5.0 (median 32.7 months). With a median TFR follow-up time of 39.3 months, 88.1% remained in TFR; 11 patients relapsed. The Kaplan-Meier probability of molecular relapse-free survival (MRFS) was 90.0% (95% CI: 89.7%-90.3%) by 12 months, and 88.7% (95% CI: 85.5%-91.9%) by both 24 and 36 months. The median halving time was longer in relapse patients (20.6 vs. 13.8 days, P=0.058). Seven patients had TKI withdrawal syndromes, including joint pain, bone pain, malaise and hypertension. The median time to regain MMR, MR4.0, and MR4.5 was 2.0 months (1.0-6.3 months), 2.5 months (1.5-12.5 months) and 4.8 months (2.0-22.4 months), respectively, after restarting TKIs. Univariate analysis identified halving time (>18.5 days vs. ≤18.5 days, HR 3.7, P=0.037), MR4.5 duration (≤56.8 months vs. >56.8 months, HR 4.9, P=0.009), and TKI treatment duration (≤77.5 months vs. >77.5 months, HR 5.1, P=0.007) as predictors for TFR failure. Multivariate analysis confirmed halving time >18.5 days (HR 3.8, P=0.033) as the strongest independent predictor. MRFS at 12 months was significantly lower in patients with halving time >18.5 days (80.4%) compared to ≤18.5 days (93.4%, P=0.025). ASXL1 mutation was present in 9 of 101 patients at diagnosis but only one remained positive at TFR initiation, he relapsed later with the mutation persisting despite regaining MR4.5 after TKI restarted. Six patients had sustained TFR after TKI switch due to prior treatment failure, 3 were ASXL1-positive compared to non-failure patients (3/6 vs. 6/95; P=0.009). Apart from ASLX1, only 3 of relapse patients had additional gene mutations, one acquired DNMT3A at relapse which disappeared after regained DMR and the other two had SFI and SH2B3 variants at diagnosis, time of TFR initiation and regained-DMR. Conclusions: Our results demonstrate that a TFR success rate of higher than 85% can be achieved in patients with a longer duration of TKI therapy, prolonged sustained DMR, and a shorter halving time, resulting in excellent MRFS that exceeds the 3-year MRFS rates of 50–60% reported in most published studies. Halving time emerged as the strong predictor for TFR failure. Although ASXL1 mutations were detected at diagnosis in a small subset of patients, they appear to be associated with initial treatment failure rather than TFR failure.

This content is only available as a PDF.
2025
Sign in via your Institution