Tyrosine Kinase Inhibitors Discontinuation in Chronic Myeloid Leukemia: Observational Study of 673 Patients in Italy

Background: Treatment free remission (TFR) is one of the main goals of therapy with tyrosine kinase inhibitors (TKIs) in Chronic Myeloid Leukemia (CML) patients (pts). In the last 15 years several studies analyzed the outcome of pts who discontinued TKIs demonstrated that it is safe according to the current recommendations.

Aims: to evaluate TFR in the setting of clinical practice for bringing out unmet clinical needs, optimizing already consolidated practices and obtaining increasingly personalized treatments based on the disease profile.

Methods: we proposed a retrospective and prospective observational study on pts with CML who discontinued TKIs in Italy. We performed descriptive statistics and survival analysis with Kaplan-Meier. Bayesian Model Averaging (BMA) was used to assess potential risk factors. BMA averages all data-supported models to estimate effect sizes of variables on the endpoint. The percentage of inclusion (PI) indicates how often a variable appears across models, suggesting its importance. Standard rules of thumb for interpreting this posterior probability are P<50% evidence against the effect; 50-75% weak evidence; 75-95% positive evidence; >95% strong evidence.

Results: We present data on the first 673 pts who discontinued TKIs in 43 centers. Median age at discontinuation was 59 yo (IQR:48-70). 313(46.5%) were female; 53%, 34.5% and 13.5% were low, intermediate, and high Sokal score respectively. 27.5% pts carried a b2a2 BCR-ABL1 transcript, 70.4% pts a b3a2 and 12 pts discontinued with an atypical transcript. 497 pts (74.6%) discontinued in first line, 148 pts in 2^nd line, 18 pts in 3^rd line, and 3 pts in fourth line. 387 pts (57.5%) were on treatment with imatinib at discontinuation, of which 362 pts (94.5%) frontline. 179 pts (26.6%) with nilotinib (113 frontline, 66 in 2^nd or further lines), 91 pts (13.5%) with dasatinib (22 frontline, 67 in 2^nd or further lines), 11 pts (1.6%) with bosutinib and 5 pts discontinued ponatinib in 2^nd or further lines.

Median duration of treatment with the last TKI was 82 mos (IQR:56-118); median time to DMR (undetectable transcript or MR4/MR4.5/MR5) with the last TKI was 17.5 mos (IR 7.8-35). Median duration of DMR was 63 mos (IQR:39-91) before suspension. 620 pts had a response defined according to molecular standardization: 6 pts (0.9%) were less than MMR, 29 pts (4.7%) were MR3, 181 pts (29.2%) were MR4, 223 pts (36%) were MR4.5, 181 pts (29.2%) were MR5. Reasons for discontinuation were toxicity for 117 (17.8%) of pts, pregnancy for 33 (5%), shared decision between clinician and patient for 440 (66.9%), other reasons for 68 pts (10.3%). Median time to restart treatment was 5.3 mos (IQR: 3.7-10).

187 pts had to restart therapy, 118 (63.1%) were treated with imatinib, 39 (20.9%) with nilotinib, 21 (11.2%) with dasatinib, 7(3.7%) with bosutinib, 2 pts (1.1%) with ponatinib. 160 pts (85.6%) restarted treatment with the same TKI that was ongoing before discontinuation. One patient restarted treatment with asciminib. One patient progressed in accelerated phase and another one patient had a lymphoid blastic crisis.

We assessed age at discontinuation, sex, Sokal score, ELTs risk, type of transcript, duration of TKI therapy, line of therapy at stop, last TKI generation, depth of MR, reasons for stop as potential prognostic factors for TFR. We have identified as risk factors associated for restarting treatment the last TKI (2nd or 3rd gen vs 1st gen HR = 0.51, P>99%), it means that those who discontinue a 2^nd or 3^rd gen TKIs have a 49% reduction of the risk of resuming treatment, regardless of the line of therapy; the duration of DMR (1% risk reduction for any additional month, HR = 0.99, P>99%), level of MR at stop (PI=38%) (MR4.5 vs MR4, HR:0.51, P = 99%, MR5 vs MR4, HR:0.52, P=99%), the duration of the TKIs therapy before discontinuation (PI=44%) with HR:0.99, meaning that for each additional month there is a risk reduction of 1%, P=97%).

Survival analysis was conducted on 655 patients. 187 pts (28.5%) had to resume therapy, most of them (N=151) lost molecular response within 12 months of stopping treatment.

Conclusions: Respect to our previous study on 283 pts, we confirm the importance of treatment duration and of the type of TKI as prognostic factors for TFR maintenance. For the first time 2 pts progressed, suggesting that we need to explore other alerts aimed at early detection of those who, despite meeting the criteria for TFR, have an increased risk of resuming treatment.