Predictors of successful tirosine kinase inhibitor discontinuation in chronic myeloid leukemia: Long-term, real-world evidence on the role of faster molecular decline and duration of response

Abstract Tyrosine kinase inhibitor (TKI) discontinuation, also known as treatment-free remission (TFR) is currently one of the main goals of chronic phase chronic myeloid leukemia (CP-CML) therapy. Approximately half of patients are eligible for TFR, but only 50% maintain remission. Key factors associated with successful TFR include the TKI duration, BCR::ABL1 transcript type, and baseline blasts percentage. However, additional real-world clinical data are needed to validate these predictors. This study assessed the success rate of TFR in CP-CML patients in real-world clinical settings. It aimed to identify new clinical criteria that could improve the maintenance of TFR in these patients. We conducted a retrospective and prospective observational study of Korean CP-CML patients who discontinued first-line TKIs after maintaining MR4.5 (BCR::ABL1 IS ≤0.0032%) for ≥4 years. Reverse transcriptase quantitative polymerase chain reaction (RT-qPCR) was used to monitor molecular response after discontinuation. Upon major molecular response (MMR) loss, the prior TKI was resumed at the same dose, with monthly RT-qPCR monitoring until MMR was regained. Altogether 306 patients (139 males and 167 females) discontinued TKI. Median age at diagnosis was 44.1 years (range, 15-79). EURO risk scores were low in 132 (43.1%), intermediate in 105 (34.3%), high in 26 (8.5%), and unknown in 43 (14.1%) patients. Prior to discontinuation, all patients received TKIs for a median of 98.1 months (range, 32.2-213.4), and maintained MR4.5 for a median 53 months (range, 25.7-118.5). At discontinuation, 223 patients (71.1%) were on first-line TKI (169 Imatinib, 24 Nilotinib, 17 Radotinib, and 13 Dasatinib), while 45 (14.7 %) and 38 (14.2%) patients were receiving their second and more than three TKIs, respectively. Median BCR::ABL1 halving time after TKI initiation was median 13.6 days (range, 6.8-57.9), reflecting early molecular kinetics. After a median follow-up from TKI discontinuation of 49.4 months (range, 18.8-155.3), TFR rates at 12, 24, and 48 months were 77.6%, 72.2%, and 70.3%, respectively. 109 patients maintained MR4.5 off treatment. Despite 197 patients losing undetectable minimal residual disease (UMRD), 109 patients regained MMR spontaneously. While 88 (28.8%) patients have lost MMR at a median of 3.7 months (range, 0.9-122.3) after discontinuation. 88 patients who lost MMR were retreated with same TKI, 83 patients (94.3%) re-achieved MMR with a median of 4.5 months (range, 1.8-23.4). Most molecular relapses occurred within six months after TKI discontinuation. However, six patients experienced late relapses: two patients after two years, three patients after three years, and one patient after ten years. The two patients with the latest molecular relapse showed recurrence at 53.2 and 122.3 months, respectively, both of which progressed to the blast phase. Notably, these patients had sustained Complete molecular response (CMR) for 50.6 and 116.2 months after discontinuation, and lost MMR within three and six months after UMRD loss. One patient is currently maintaining CMR following allogeneic stem cell transplantation, while the other died during treatment. Among all patients with TFR failure, 26 (29.5%) attempted a second TFR. Of those who experienced a second TFR failure, 12 (66.7%) proceeded to a third TFR, and among patients who failed a third TFR, 2 (22.2%) attempted a fourth TFR. The success rates of each TFR attempt were 30.8%, 25%, and 0%, respectively. Multivariate analysis showed molecular relapse was significantly associated with BCR::ABL1 halving time (<13.6 days vs. ≥13.6 days, HR 2.395, P=.034), duration of CMR (<24.8 months vs. ≥24.8 months, HR 0.262, P=.003), EURO risk score (Low vs. Intermediate vs. High, HR 9.900, 21.854, P<0.001, P=0.010), withdrawal symptoms (HR 0.401, P<0.001), and age at TKI discontinuation (<52.4 years vs. ≥52.4 years, HR 0.107, P=.005). Our results demonstrate that patients who receive TKI treatment for approximately 9 years and maintain a DMR for more than 5 years may achieve successful TFR, regardless of the TKI type. Late molecular relapse beyond two years occurred in 6.8% of all relapsing patients, underscoring the need for long-term monitoring even in patients with durable TFR. The study also highlights the importance of identifying predictive factors such as halving time and duration of CMR for sustained TFR. In addition, multiple attempts at TFR may be achievable in selected patients.