Systematic Literature Review of the Efficacy of Available Interventions for Patients with Chronic-Phase (CP) Chronic Myeloid Leukemia (CML) Who Have Previously Been Treated with >= 2 Lines of Tyrosine Kinase Inhibitors (TKIs)

Context and Objective Despite several TKIs being available as standard of care for CP-CML patients, survival rates are low once patients move beyond their second line of treatment. This study aimed to systematically capture and summarize evidence on treatment outcomes for CP-CML patients previously treated with ≥2 lines of TKIs. Methods The search was conducted in November 2019 within Embase, MEDLINE, and Cochrane, to identify studies of TKIs, allogeneic stem cell transplantation (allo-SCT), omacetaxine, and hydroxycarbamide in the target population. Outcomes of interest were complete cytogenetic response (CCyR), major molecular response (MMR), as well as 4-log (MR4), 4.5-log (MR4.5), and 5-log (MR5) reductions in BCR-ABL ratio, and survival (progression-free [PFS] and overall [OS]). Results Seven single-arm trials and 12 observational studies were included, with no randomized controlled trial (RCT) of relevance being identified. Six studies featured multiple interventions, all of which were observational. Seven studies evaluated dasatinib, seven nilotinib, six bosutinib, five ponatinib, one omacetaxine, one imatinib, and one allo-SCT. Sample sizes ranged between 10 and 270 patients. Median age and proportion of males ranged from 33 to 68 years and from 45% to 65%, respectively. Among the seven studies reporting on performance status, patients had an ECOG score of either 0 or 1. Between 0% and 54% of patients were in CCyR at study entry. Median PFS and median OS were not reached with ponatinib, bosutinib, dasatinib, or nilotinib, were not reported with allo-SCT, and were 9.6 and 33.9 months with omacetaxine. CCyR, MMR, MR4, and MR4.5 rates (reported in 18, 18, five, and five studies respectively) ranged from 5% to 81%, 8% to 72%, 20% to 57%, and 16% to 57%. No study reported MR5. Conclusions No RCT has been conducted in this population. Since most included studies were observational, with known associated risks of bias, there is a clear need for data generation through RCTs. Long-term outcomes were generally poor; differences in prognostic factors at baseline may explain part of the heterogeneity in the reported outcomes. Comparative evidence on the efficacy of interventions is needed to aid clinicians in managing this heavily pretreated population.