"Duration of Deep Molecular Response" Has Most Impact on the Success of Cessation of Tyrosine Kinase Inhibitor Treatment in Chronic Myeloid Leukemia - Results from the EURO-SKI Trial

Background: The advent of tyrosine kinase inhibitors (TKI) has decisively improved the survival of patients with chronic myeloid leukemia (CML). Many patients achieve deep molecular response (DMR), a prerequisite for cessation of TKI therapy. The goal of the EURO-SKI trial was to assess the safety and feasibility and to define precise conditions for stopping treatment.

Methods: Chronic-phase CML patients without prior failure to any TKI, treated for at least three years and in DMR for at least one year were eligible for stopping TKI. DMR was defined as MR⁴ (detectable BCR-ABL1 on International Scale (IS) ≤ 0.01% or undetectable BCR-ABL1 in samples with ≥ 10,000 ABL1 or ≥ 24,000 GUS transcripts, respectively). Molecular responses were assessed by real-time quantitative PCR (RQ-PCR) at designated standardized laboratories. Primary outcome was molecular recurrence- (MRecFS) and treatment-free survival (MRecTFS) defined by staying at least in major molecular remission (MMR, BCR-ABL (IS) < 0,1%) without restart of therapy. MRecTFS was estimated with the Kaplan-Meier method.

For the prognostic logistic regression analysis of the endpoint "MRecFS status after 6 months", candidate variables were age at diagnosis and at TKI discontinuation, sex, Sokal, Euro, EUTOS, and ELTS risk scores, and, alternatively, any variable part of any of the scores, previous IFN treatment, duration of TKI treatment, time to DMR while under TKI treatment, and DMR duration until discontinuation of TKI. Prognostic results were examined in an independent validation sample.

Results: 821 patients (48% female, median age 60 years at study entry, range 19-90) were recruited, 755 patients were assessable for the estimation of MRecTFS. 371 (49.1%) lost MMR after TKI cessation, 13 restarted TKI in MMR, and 4 died in MMR. MRecTFS was 60% at 6 months (95% confidence interval (CI): 56-63%) and 49% at 24 months (CI: 45-52%). For 224 (29.7%) patients a grade 1-2 and for 9 patients a grade 3 TKI withdrawal syndrome was reported.

After treatment restart, so far 321 (86%) and 302 (81%) of 373 patients re-achieved MMR and MR⁴, respectively. Median time to regain MMR and MR⁴ was 3 and 4 months, respectively.

In a prognostic model, of 448 patients (186 relapses after 6 months, 42%), 77 (17%) received IFN pre-treatment before imatinib was started. As IFN-pretreatment >1.5 years had significant positive influence on MRecFS, these patients were omitted in the prognostic model. In 405 imatinib treated patients TKI-treatment duration and DMR duration were significantly (p<0.005) correlated with MRecFS at 6 months. The minimal p-value approach identified a threshold of 5.8 years for imatinib-therapy duration and 3.1 years for DMR duration distinguishing two groups with 17% difference in MRecFS at 6 months. However, as MR⁴ duration before TKI discontinuation was part of duration of imatinib treatment, both parameters were strongly correlated. The influence of the variable "duration of imatinib treatment before MR⁴ achievement" was not significant (p=0.50).

Considering all candidate variables including the new variable "duration of imatinib treatment before MR⁴ achievement" for multivariate modelling, MR⁴ duration and duration of imatinib treatment before MR⁴ achievement were weakly negatively correlated (Spearman's rank correlation coefficient: -0.29), and only MR⁴ duration remained significant (p=0.0009; duration of treatment before MR⁴: p=0.0921

The validation sample consisted of 195 patients (73 relapses after 6 months, 37%). MR⁴ duration had the same odds ratio of 1.13 (CI: 0.98-1.29) as in the learning sample but were not significant anymore (p=0.08).

Conclusion:

EURO-SKI outlines important preconditions which can be employed as guidance for stopping criteria. Criteria with improved impact for successful stopping could be defined using a prognostic modeling to predict MRecFS at 6 months in a large cohort of patients who stopped imatinib. Best cut-off was defined by 3.1 years for MR⁴ duration and 5.8 years for imatinib-therapy duration whereas the later was less important.

Apart from the two cut-offs, an almost linear increase in MRecFS probability per additional year on treatment / in MR⁴ for patients treated first-line with imatinib could be demonstrated. In conclusion, for imatinib therapy, a minimal total treatment duration of 6-years is suggested of which the patient should have been in MR4 for at least 3 years before stopping.