Abstract
Introduction: First- and second-generation (1G and 2G) tyrosine kinase inhibitors (TKIs) have dramatically improved the outcomes of CML-CP, transforming it into a manageable chronic condition, with some patients achieving a near-normal life expectancy. Consequently, managing TKI side effects is key to maintaining adherence and response. Asciminib, the first BCR::ABL1 inhibitor to specifically target the ABL myristoyl pocket, has demonstrated superior efficacy and favorable tolerability against investigator-selected (IS) 1G and 2G TKIs (ASC4FIRST, [NCT04971226]) in newly diagnosed CML-CP. However, ASC4FIRST was not powered to show statistical significance in direct comparisons between individual 2G TKIs and asciminib. This study aimed to indirectly compare the efficacy of dasatinib and asciminib across randomized controlled trials (RCTs) using matching-adjusted indirect comparison (MAIC) to account for heterogeneous baseline characteristics between and within the RCTs.
Methods: To inform the MAIC, a systematic literature review (through September 28th, 2023) identified all Phase III RCTs with (1) adults newly diagnosed with CML-CP treated with asciminib (ASC) or dasatinib (DAS); (2) and assessed for clinical efficacy outcomes (i.e., MR2, major molecular response [MMR], MR4, MR4.5) by 48 and 96 weeks. Two trials were identified (ASC4FIRST and DASISION [NCT00481247]). Trial endpoints varied in timing (ASC4FIRST: 48 and 96 weeks, DASISION: 12 and 24 months), and DASISION utilized complete cytogenetic response rather than MR2. Patients were grouped into ASC and DAS cohorts by assigned treatment, irrespective of the trial in which they participated. To adjust for baseline cohort differences in relevant treatment effect modifiers and prognostic factors, a logistic model was built to predict treatment based on demographics (age, sex, race), clinical characteristics (WBC, platelet count), and severity at diagnosis (ECOG PS). Risk score adjustment was not feasible due to differing systems across trials. ASC patients were individually weighted via the logistic model such that their average characteristics were comparable to the DAS cohort's, which were calculated as the weighted average of trial level DAS characteristics, weighted by trial patient count. DAS cohort outcomes were pooled across trials using inverse-variance weighted meta-analysis. All patients randomized to ASC in ASC4FIRST were included regardless of treatment strata (i.e., the comparator IS-TKI specified prior to randomization). Relevant limitations include heterogeneity between studies, including differences in risk score systems, endpoint timelines, evolved BCR::ABL1 assay methodologies and standardization, and the evolving landscape of approved therapies for CML during the study periods. Estimated weighted odds ratios (OR) and 95% confidence intervals (95% CI) are reported for all comparison between ASC 80 mg QD and DAS 100 mg QD.
Results: ASC cohort included 200 patients (Effective Sample Size =142.01 after weighting), and DAS cohort included 301 patients. After weighting, characteristics were similar across cohorts, including age (mean±sd: ASC 46.50±14.92 vs DAS 46.5±14.92), proportion male (54.82% vs 54.82%), race (white: 55.13% vs 50.83%, black: 1.56% vs 0.7%, Asian: 42.52% vs 42.5%, other: 0.78% vs 5.9%), WBC ≤ 25.1% (47.2% vs 47.2%) platelet count ≤ 448 (48.5% vs 48.5%), and ECOG PS = 0 (81.73% vs 81.7%).
By 48 weeks, patients in the ASC cohort had statistically significantly higher rates of MMR (OR [95% CI]: 2.58 [1.67, 3.98]), MR4 (3.64 [2.29, 5.79]), and MR4.5 (3.54 [1.94, 6.46]) than patients in the DAS cohort. This trend continued by 96 weeks (MMR: 2.16 [1.33, 3.51]; MR4: 2.14 [1.42, 3.23]; MR4.5: 2.11 [1.35, 3.30]). Rates of MR2 were numerically superior in the ASC cohort by 48 weeks (1.94 [0.95, 3.97]) and 96 weeks (1.94 [0.95, 3.97]).Conclusions: After controlling for available treatment effect modifiers and prognostic factors, asciminib patients experience significantly improved efficacy versus patients on dasatinib.
