AUTHORS
Omer Jamy, MD, Rodney Sparapani2, Ryan Conrardy3, Ravi Bhatia, MD4, Anthony Michael Hunter, MD5, Arielle Baim6, Heidi Ortiz7, Emily Giever8, Kathryn Flynn, PhD9, Alexis Visotcky10, Jorge Cortes, MD11, Richard Larson12, B. Douglas Smith, MD13, Kendra L. Sweet, MD, MS14, Aaron Winn15, Ehab Atallah, MD16, James Thompson, M.D., M.S.17
AFFILIATIONS
1 The University of Alabama at Birmingham, Birmingham, Alabama, United States
2 Medical College of Wisconsin, Milwaukee, United States
3 Medical College of Wisconsin, Milwaukee, United States
4 The University of Alabama at Birmingham, Birmingham, Alabama, United States
5 Emory University, Atlanta, United States
6 Medical College of Wisconsin, Milwaukee, United States
7 Medical College of Wisconsin, Milwaukee, United States
8 Medical College of Wisconsin, Milwaukee, United States
9 Medical College of Wisconsin, Milwaukee, United States
10 Medical College of Wisconsin, Milwaukee, United States
11 The University of Alabama at Birmingham, Birmingham, Alabama, United States
12 The University of Chicago, Chicago, United States
13 Johns Hopkins University, Baltimore, United States
14 Moffitt Cancer Center, Tampa, United States
15 The University of Illinois, Chicago, United States
16 Medical College of Wisconsin, Milwaukee, United States
17 Roswell Park, Buffalo, United States
ABSTRACT
Background
Treatment-free remission (TFR) represents an important therapeutic goal for patients with chronic myeloid leukemia (CML) who achieve sustained deep molecular response (DMR). Although prospective clinical trials have established the feasibility of TFR in selected patients, real-world data describing patterns of TFR attempts and outcomes after discontinuation remain limited. We evaluated TFR practices and long-term outcomes in a multicenter real-world CML cohort.
Methods
Patients with CML treated across six academic centers were retrospectively identified. Demographics, treatment history and survival outcomes were collected. TFR was defined as an attempted discontinuation of TKI therapy after achieving response criteria according to published guidelines. Time-to-death survival analysis was performed with a nonparametric machine-learning approach based on Bayesian Additive Regression Trees (BART). TFR status was a time-dependent covariate to account for changes in TFR status over the course of follow-up. Additional covariates included sex, race, age at diagnosis, TKI treatment line, TKI choice and TKI dose intensity. The BART survival model evaluated the association between these factors and the risk of death allowing for complex nonlinear relationships/interactions among predictors.
Results
There were 412 patients in the database at the time of the analysis with a median follow-up of 8 years (range 0.1-34). The median age at diagnosis was 51 years (range 18–89), 51% were male and 81% were White. Initial TKI therapy included imatinib (51%), dasatinib (35%), nilotinib (9.5%), and other TKIs (4.5%); 44% of patients underwent at least one TKI change. Patients were on their 1st (31%), 2nd (30%) or ≥3rd (38%) TKI. Current TKI therapy included dasatinib (34%), imatinib (20%), asciminib (19%), bosutinib (11%) and ponatinib (9%). Common comorbidities included hypertension (37%), hyperlipidemia (23%), diabetes mellitus (13%), GERD (13%), obesity (13%) and coronary artery disease (8.4%). Overall, best molecular response was BCR::ABL1 <0.0032% in 62% of patients, BCR::ABL1 <0.01% in 6.6% of patients and BCR::ABL1 <0.1% in 15% of patients.
A total of 67 (16%) patients attempted TFR. The median age at TKI discontinuation was 62 years (range 49-68), 61% were female and 88% were White. The median time from CML diagnosis to TFR attempt was 8 years (IQR 5-11.5) and the median number of TKIs before TFR was 2 (range 1-5). TKI therapy at the time of discontinuation included imatinib (40%), dasatinib (25%), nilotinib (16%), bosutinib (9%), ponatinib (6%), and asciminib (3%). Following discontinuation, 52 patients (78%) remained off therapy, whereas 14 patients (21%) restarted TKI therapy and one patient had unknown status. The median time to loss of TFR was 6 months (range 2-46) with 2 patients losing TFR ≥2 years. Of the 14 patients who restarted TKI, 11 achieved DMR again at last follow up. Among patients remaining off therapy, median TFR duration was 2.9 years (range 0.1–14.9).
Overall, 70 patients died (67=never attempted TFR, 3=attempted TFR). Causes of death were predominantly unrelated to CML. The time-to-death survival analysis demonstrated differences in OS according to the timing of first TFR attempt. Earlier first TFR attempts were associated with more favorable observed overall survival. Estimated OS at 20 years was 91%, 88%, 81%, and 70% among patients with first TFR attempts after 3.1 years, 4.2 years, 9.0 years, and those who never attempted TFR, respectively.
Conclusions
In this multicenter real-world CML cohort, TFR was achievable in a subset of patients who attained DMR, with three-quarters remaining off therapy after discontinuation. Earlier TFR attempts were associated with favorable long-term outcomes and may represent a marker of favorable disease biology characterized by rapid achievement of durable molecular control and lower dependence on ongoing TKI therapy. These findings support TFR as an important therapeutic goal in CML while emphasizing that associations between TFR timing and survival should be interpreted with caution given potential selection bias and residual confounding inherent to observational analyses. Further studies are needed to validate these
DISCLOSURES
Omer Jamy, MD: Ascentage - Consultancy (Includes expert testimony); SELLAS - Consultancy (Includes expert testimony); TERNS - Consultancy (Includes expert testimony)