Introduction: Several tyrosine kinase inhibitors (TKIs) are now available for the management of patients with chronic myeloid leukemia (CML). Although clinical trials (CTs) have established the efficacy and safety of TKIs, the generalizability of these findings to broader real-world populations remains controversial.

Methods: We identified CTs supporting FDA approval of TKIs for newly diagnosed CML by reviewing FDA databases (Oncology/Hematologic Malignancies Approval Notifications and Novel Drug Approvals Databases). Associated manuscripts were retrieved from PubMed. Real-world data (RWD) were obtained via a systematic review of observational studies across five major biomedical literature databases. For uniformity in this analysis, we focused on imatinib trials (including the imatinib arm of randomized trials) and RWD only. Differences between cohorts were assessed using unpaired t-tests for means and Chi-square tests for proportions.

Results: We included the imatinib arm of 5 pivotal CTs: the IRIS, which established the superiority of imatinib over interferon alfa plus cytarabine, and the DASISION, ENESTnd, BFORE, and ASC4FIRST trials, which supported the approvals of dasatinib, nilotinib, bosutinib, and asciminib, respectively. The RWD cohort included 20 manuscripts. A total of 8,788 patients were analyzed: 1,437 from CTs and 7,351 from RWD. The mean sample sizes per study were comparable (CTs: 287.4 ± 164.2; RWD: 367.5 ± 422.4; p 0.52). Sex distribution was similar with a male predominance (CTs: 60%, RWD: 58.2%; p=0.11), though patients in the RWD were significantly older (51.3 ± 4.5 years vs.49.8 ± 3 years; p<0.01). Black patients were markedly underrepresented in CTs vs. RWD (2.9% vs.15.7%; p<0.01), while Asian patients were overrepresented (21.6% vs.6.1%; p<0.01). Similarly, Hispanic patients accounted for only 2% of CT participants, compared to 16.6% in the RWD cohort (p<0.01). Notably, Africa and South America were underrepresented in CTs and RWD, while Asia, Europe, North America, and Oceania were predominantly represented in both cohorts. RWD patients had a significantly longer interval from diagnosis to treatment initiation (19.9 vs.0.8 months; p<0.01). More CT patients were classified as low Sokal risk (49.6% vs.38.4%), while the RWD group had a higher proportion of intermediate-risk patients (36.9% vs.25.7%; p<0.01). Splenomegaly was more prevalent among RWD (64.9% vs.19.9%; p<0.01).

Follow-up duration was nearly twice as long in CTs (96 vs.47.4 months; p<0.01), and treatment outcomes were consistently better in this group. Rates of early molecular response (BCR::ABL1 ≤10% at 3 months) were significantly higher in CTs (62.7% vs.23.9%), as was complete cytogenetic response at 6 months (52.3% vs.33.2%) and 12 months (96.6% vs.32.1%) (All p<0.01). Major molecular response was also more commonly achieved in the CTs at both 12 months (38.7% vs.32.9%; p<0.01) and 24 months (48.9% vs.43.8%; p=0.01).

The rate of progression to advanced phases of CML was comparable between the cohorts (CTs: 5.6% vs. RWD: 4.9%; p=0.39), as well as progression-free survival at 5 years (88.7% vs.88.4%; p=0.85). All-cause mortality was significantly lower in the real-world cohort (4.9% vs.9.5%; p<0.01), and a similar trend was observed for CML-related mortality (3.2% vs.4.7%, p=0.05;). However, overall survival at 5 years remained equivalent between groups, with survival rates exceeding 90% (CTs: 90.9% vs.RWD: 90.6%, p=0.85). Treatment discontinuation rates were significantly higher in CTs due to both intolerance (7.7% vs.4.1%; p<0.01) and other non-efficacy-related causes (12.6% vs.3.2%; p<0.01). In contrast, discontinuation due to suboptimal efficacy was more frequent in the RW cohort (15.6% vs.23.7%; p<0.01).

Conclusions: Patients in the RWD cohort exhibited higher clinical risk profiles, greater racial and ethnic diversity, and longer intervals between diagnosis and treatment initiation. These factors may contribute to the lower rates of achieving optimal treatment responses, including cytogenetic and molecular milestones. Despite these differences, long-term outcomes were comparable, aligning with current trends showing that survival in CML patients now approaches that of the general population. Our findings underscore the partial generalizability of CT data and highlight the need to enhance the inclusion of diverse and high-risk populations in future studies to better inform real-world clinical decision-making.

This content is only available as a PDF.
2025
Sign in via your Institution