The advent of tyrosine kinase inhibitor (TKI) has revolutionized the treatment of chronic myeloid leukemia (CML) and improved the prognosis of patients to near-normal levels (Sasaki et al., 2015). However, the adverse effects associated with long-term TKI therapy are an overwhelming factor affecting the quality of life (Steegmann et al., 2016). Besides, the high cost of persistent TKI treatment is increasingly becoming an important issue (Phuar et al., 2019). Over recent years, numerous clinical studies have proved that about 40%–60% of patients with deep molecular response (DMR) could achieve TFR (Rea et al., 2017b; Etienne et al., 2017; Mahon et al., 2018; Saussele et al., 2018; Nagafuji et al., 2019; Kimura et al., 2020; Shah et al., 2020). In this context, treatment-free remission (TFR) is a major goal for all CML patients according to the current guidelines (Hochhaus et al., 2020). Recently, The Gruppo Italiano Malattie EMatologiche dell’Adulto (GIMEMA; Italian Group for Hematologic Diseases of the Adult) CML Working Party developed a project designed to select treatment policies that might increase the likelihood of achieving TFR. A consensus was reached on disease risk assessment, first-line treatment, and the responses that require a change of the TKI, contributing to optimizing the treatment strategy for TFR (Baccarani et al., 2019). However, data on the impact and outcomes of TKI discontinuation outside of clinical studies remain sparse. In clinical practice, TKI dose reduction is also an important aspect of alleviating adverse effects and improving quality of life (Russo et al., 2015; Iurlo et al., 2020). In addition, several studies have shown that low-dose TKI treatment is effective in maintaining major molecular response (MMR) (Rea et al., 2017a; Clark et al., 2017; Claudiani et al., 2021). Moreover, dose-halving therapy yields higher molecular recurrence-free survival rate compared to direct TKI discontinuation (Luo et al., 2022). Our recent study also confirmed the feasibility of TKI dose reduction therapy for Chinese CML patients (Chen et al., 2022). The concern of whether dose reduction before TKI discontinuation affects patient’s access to TFR is also of interest to investigators. The DESTINY study has investigated the effects of low-dose TKI treatment withdrawal before stopping TKI, of which patients with MMR at dose reduction experienced a TFR rate of 36% at 36 months and patients with DMR reported a TFR rate of 72% (Clark et al., 2019), indicating TKI de-escalation is safe for most patients with DMR. TKI de-escalation - to discontinue TKI may be an alternative way to achieve TFR (Kunbaz and Eskazan, 2019). Still, more studies are needed to investigate the possibility of TKI discontinuation in patients with dose reduction, prior treatment failure, or other problems in the real world. In this study, we retrospectively analyzed a real-world series of patients with CML who discontinued TKI, including patients who received low-dose TKI therapy prior to discontinuation.

Currently, long-term TKI treatment is no longer indicated for all patients with CML, as some patients who obtain DMR can safely discontinue TKI. For example, the DASFREE study showed the 1-year TFR rate was 48%, which was 58% in the ENESTop study (Mahon et al., 2018; Shah et al., 2020). The EURO-SKI study with the largest number of patients reported a TFR rate of 61% at 6 months, 50% at 2 years, and 49% at 3 years (Saussele et al., 2018). In addition, several results from real-world studies demonstrated the feasibility of TKI discontinuation. An observational study reported an estimated TFR at 12 months of 69% for Italian patients (Fava et al., 2019). A study from a Spanish research group described a TFR rate at 4 years of 64% (Hernandez-Boluda et al., 2018). Recently, a Swedish discontinuation study showed that 62.2% of patients maintained TFR at the last follow-up (Flygt et al., 2021). However, limited data are available on TKI discontinuation for Chinese CML patients. In the present study, a total of 190 patients who discontinued TKI medication were included in this study. The estimated TFR was 68.8%, and 65.5% at 12 and 24 months, comparing favorably to the TFR rate of approximately 50% in most clinical studies (Saussele et al., 2018). The rate is also similar to the TFR rate reported in Italian, Spanish, and Swedish patients (Hernandez-Boluda et al., 2018; Fava et al., 2019; Flygt et al., 2021). After discontinuation, most relapse occurred within the first 6 months. In this study, 72.4% of patients lost MMR within 6 months, and 93.1% of patients lost MMR within 12 months. One patient lost MMR at 23 months after TKI discontinuation. A recent study showed an estimated rate of molecular recurrence after 2 years of discontinuing imatinib of 18% (Rousselot et al., 2020). Hence, prolonged molecular monitoring is still necessary for patients who have not relapsed in the early phase of drug discontinuation. Most patients (56/57) who were restarted TKI regained MMR, and 47 patients regained MR4 with a median time of 4 months (range, 1–11). The results within the Chinese population showed that TKI discontinuation is safe and feasible also outside controlled clinical trials.

At present, dose optimization is increasingly emphasized as an important part of individualized treatment for CML patients with different demands. In the present study, 27 patients experienced dose reduction before stopping TKI for adverse events (n = 17), desire to stop TKI (n = 8), and finical burden (n = 2). Patients with low-dose TKI treatment had a TFR of 63.8% at 6 months and 55.8% at 12 months. Similarly, patients with full-dose displayed a TFR of 78.4% and 70.3% at 6 months and 12 months, respectively. In the study of Cayssials et al. (2020), the TFR rate at 12 months was 56.8% for full-dose and 80.8% for the low-dose group. However, the DESTINY study reported patients experiencing dose reduction had a TFR rate of 72% after 24 months off TKI (Clark et al., 2019), which appeared to be preferable to the TFR rate of 50% in the EURO-SKI study (Saussele et al., 2018). Claudiani et al. (2021) reported a TFR of 74.1% among patients who have experienced dose reduction. In Claudiani et al. (2021) study, the median duration of MR4 before stopping TKI was 6.1 years, significantly longer than that in the EURO-SKI of 3.1 years. Thus, they suggested that the increased TFR in the low-dose group might be related to the prolonged duration of MR4 to stopping TKI. Recently, Iurlo et al. (2022) investigated the effect of TKI reduction on TFR in a large multicenter cohort of 194 patients with low-dose TKI. Iurlo et al. (2022) study, 71.1% patients were still in TFR after a median follow-up of 29.2 months, which is better than the 62% reported in a recent Italian study with 293 patients who discontinued TKI after a median follow-up of 34 months (Fava et al., 2019; Iurlo et al., 2022). Still, it remains controversial whether the dose reduction before TKI withdrawal can lead to an improvement in TFR. 81.5% (22/27) low-dose patients experienced half of the standard-dose treatment. Moreover, we found the degree of TKI dose reduction was not related to the successful discontinuation, in agreement with the study of Cayssials et al. (2020). Fassoni et al. (2018) developed a mathematical model based on selected patients in the IRIS and the CML-IV study, suggesting that a TKI dose reduction of at least 50% does not affect long-term treatment. In addition, we found that the duration of dose reduction also did not affect the acquisition of TFR. Interestingly, Iurlo et al. (2022) found TFR was significantly better after dose reduction due to adverse events than those with dose de-escalation after DMR achievement. Consistent with the literature, our data showed that dose reductions do not hinder the achievement of TFR, and TFR might be independent of the degree of reduction of TKI.

In the univariate analysis, the type of TKI, TKI resistance, line of therapy, and time to MR4, was the clinical variable significantly associated with molecular recurrence. Multivariate analysis showed patients with TKI resistance presented with significantly lower TFR rates than those without resistance. Similarly, imatinib resistance is an important risk factor for molecular relapse in the DADI trial (Okada et al., 2018). In ENESTnd 10-year analysis, patients treated with first-line nilotinib achieved a higher rate of TFR eligibility compared to first-line imatinib (Kantarjian et al., 2021). A previous study also showed that first-line 2-3G TKIs compared to imatinib were significantly associated with a better TFR (Etienne et al., 2020). In the present study, we compared the TFR of patients treated with imatinib continuously and patients transferred to 2G-TKI from imatinib. We found patients transferred to 2G-TKI treatment for imatinib resistance or non-resistance reasons had remarkably higher TFR rates than those who did not. In addition, patients who switched to 2G-TKI therapy for imatinib resistance or non-resistance reasons achieved similar TFR rates compared to patients treated with 2G-TKI in the first line. Likewise, patients treated with nilotinib as first-line and second-line experienced a similar TFR rate (58% vs. 51.6%) at 48 weeks in the ENESTfreedom study and the ENESTop study (Mahon et al., 2018; Radich et al., 2021). A recent study showed that patients treated with 2G-TKI as first-line treatment and at discontinuation had a lower net probability of TKI re-initiation (p = 0·017) compared with patients treated with imatinib (Flygt et al., 2021). In general, our results combined with data from other studies suggested that for patients with a need for discontinuation, second-generation TKI therapy may yield higher TFR rates than imatinib, especially for imatinib-resistant patients.

Several studies reported some patients developed or worsened musculoskeletal pain after discontinuing TKI, suggesting the presence of withdrawal syndrome. For example, the KID study reported that 30% of patients underwent withdrawal syndrome after TKI withdrawal (Lee et al., 2016). Berger et al. (2019) analyzed the withdrawal syndrome after stopping TKIs by combining the patients included in the STIM2 and EURO-SKI trials and showed that 23.2% of patients developed withdrawal syndrome. In addition, 21% of patients in the DESTINY study reported new musculoskeletal symptoms during the dose-reduction phase (Clark et al., 2017). In the present study, 14 patients (7.4%) developed transient musculoskeletal pain within 12 months after TKI discontinuation. Thirteen patients reported a grade 1–2 TKI withdrawal syndrome and one patient reported a grade 3 TKI withdrawal syndrome. Three patients lost MMR after the occurrence of musculoskeletal pain.

In conclusion, this retrospective study showed TKI discontinuation in the real-world is as feasible and safe as in clinical studies. Indeed, dose optimization is an important consideration in individualized treatment for patients with different demands. Moreover, we found TFR rates were not impaired in patients experiencing dose reduction before TKI discontinuation compared to patients with full-dose TKI. Our data may provide a basis for the safety and feasibility of dose optimization before drug withdrawal. For patients with a need for discontinuation, second-generation TKI therapy may yield higher TFR rates than imatinib, especially for imatinib-resistant patients. However, our study still has limitations as a retrospective study. In addition, only 27 patients discontinued low-dose TKI in this study, and the data on discontinuation after dose reduction may be biased. Some patients with a long history of disease did not have BCR-ABL1 mutations test, so we did not analyze the effect of BCR-ABL1 mutations on TFR. As real-world data are still increasing, more studies are needed to accumulate data in daily practice to confirm the feasibility of dose reduction.