High BCR–ABL/GUSIS levels at diagnosis of chronic phase CML are associated with unfavorable responses to standard-dose imatinib

Abstract

Purpose: The approval of second-generation tyrosine kinase inhibitors (TKIs) for the first-line treatment of chronic myeloid leukemia (CML) has generated an unmet need for baseline molecular parameters associated with inadequate imatinib responses. Experimental Design: We correlated BCR–ABL/GUSIS and BCR–ABL/ABL transcripts at diagnosis with the outcome—defined by the 2013 European LeukemiaNet recommendations—of 272 patients newly diagnosed with CML receiving imatinib 400 mg/daily. Applying receiver-operating characteristic curves, we defined BCR–ABL/GUSIS and BCR–ABL/ABL levels associated with lower probabilities of optimal response, failure-free (FFS), event-free (EFS), transformation-free (TFS), and overall survival (OS). Results: With a median follow-up of 60 months, 65.4% of patients achieved an optimal response (OR), 5.6% were classified as "warnings," 22.4% failed imatinib, and 6.6% switched to a different TKI because of drug intolerance. We recorded 19 deaths (6.9%), seven (2.5%) attributable to disease progression. We found that higher BCR–ABL/GUSIS levels at diagnosis were associated with inferior rates of OR (P < 0.001), FFS (P < 0.001), and EFS (P < 0.001). Elevated BCR–ABL/GUSIS levels were also associated with lower rates of TFS (P = 0.029) but not with OS (P = 0.132). Similarly, high BCR–ABL/ABL levels at diagnosis were associated with inferior rates of OR (P = 0.03), FFS (P = 0.001), and EFS (P = 0.005), but not with TFS (P = 0.167) or OS (P = 0.052). However, in internal validation experiments, GUS outperformed ABL in samples collected at diagnosis as the latter produced 80% misclassification rates. Conclusions: Our data suggest that high BCR–ABL transcripts at diagnosis measured using GUS as a reference gene identify patients with CML unlikely to benefit from standard-dose imatinib.