Inverted U-shaped relationship between pre-discontinuation TKI dose intensity and treatment-free remission outcome in chronic myeloid leukemia.

INTRODUCTION: Treatment-free remission (TFR) is an established goal for patients with chronic myeloid leukemia in chronic phase (CML-CP) who achieve sustained deep molecular response (DMR) on tyrosine kinase inhibitor (TKI) therapy. However, the relationship between pre-discontinuation dose intensity and TFR outcome remains poorly characterized. METHODS: We retrospectively analyzed 146 patients with CML-CP who attempted TFR at Samsung Medical Center between 2003 and 2025. Dose intensity (DI) was defined as the time-weighted average ratio of actual to standard TKI dose during the 2-year period preceding discontinuation. The association between DI and TFR failure was modeled using Cox regression with restricted cubic splines (RCS) in the DI subcohort (n=126), using a DI of 1.0 as the reference. RESULTS: In the full cohort, at median follow-up of 22.3 months, 50 patients (34.2%) experienced molecular relapse. TFR probability was 79.5% at 6 months and 61.9% at 48 months. In the DI subcohort, an unadjusted RCS model suggested a nonlinear, inverted U-shaped pattern, but the overall DI effect was not significant (likelihood ratio P = .101; P for nonlinearity = .045). In exploratory post-hoc groups, intermediate DI showed the highest TFR failure rate (49.0% vs. 25.0% and 28.6%; P = .035). The pattern did not reach significance after adjustment for TKI generation (P for nonlinearity = .074). CONCLUSION: These exploratory findings suggest that the biology underlying DMR maintenance, rather than the absolute pre-discontinuation dose, may influence TFR outcome. Given the modest sample size and post-hoc design, this hypothesis-generating observation requires confirmation in independent cohorts.