Secondary imatinib resistance occurs in a minority of patients with chronic phase (CP) CML and is mainly due to BCR-ABL mutations. We previously suggested a practical molecular approach for the selection of patients to assess for mutations based on a significant rise in the BCR-ABL level. This selection criterion limited the number of patients requiring screening. However, retrospective analysis showed that the mutations were detectable for a number of months prior to the rise in half of the patients. Therefore, further refinement of the criteria for screening may allow earlier detection and permit timely therapeutic intervention. We determined the independent variables associated with mutations in 222 CP patients. The median duration of imatinib therapy was 24 months (range 3 to 55). Mutations were detected in 32 patients by direct sequencing at a median of 9 months (range 3 to 35). The estimated probability of a mutation by 24 months was 26% for patients treated with second-line imatinib (n=91) and 7.3% for newly diagnosed patients (n=131), P=0.0005. Landmark analysis at 9 months demonstrated a significantly poorer progression free and overall survival for patients with mutations compared to those without (21% vs 92%, 55% vs 98% respectively, P<0.0001). Multivariate analysis of pre-imatinib factors identified detectable peripheral blood (PB) blasts and an interval of more than 2 years from diagnosis as being adverse factors for the detection of mutations. This analysis was used to generate low, intermediate and high risk groups with estimated mutation rates of 5.6%, 22%, and 70% at 24 months, P<0.0001. Patients at low risk had none of the adverse factors. Intermediate risk patients had detectable PB blasts of 2% or less, or were greater than 2 years since diagnosis. High risk patients had detectable PB blasts and were greater than 2 years since diagnosis. This group also included the sole adverse factor of greater than 2% PB blasts. In our patient cohort, 9% were classified as high risk. After commencement of imatinib therapy, failure to achieve a major cytogenetic response (MCR) by 6 months was also independently predictive for a mutation. Of the patients with low and intermediate risk, those who failed to achieve a MCR by 6 months had a significantly higher probability of a mutation, 48% vs 6.6% at 24 months, P<0.0001. 14% of patients failed to achieve a MCR by 6 months and were classified as low or intermediate risk. Based on these data we propose the following strategy:

  • patients classified as high risk be screened for mutations 3 monthly from the start of imatinib therapy,

  • patients with low or intermediate risk who fail to achieve a MCR by 6 months receive 3 monthly analysis from that time-point, and

  • all other patients be screened for mutations upon a significant rise in BCR-ABL.

Based on a policy of screening all patients 3 monthly, 1,605 analyses would be needed to identify the patients in our cohort with mutations at the earliest timepoint. Using the risk-based strategy, 23% of patients would qualify for 3 monthly screening and only 282 analyses undertaken. The requirement for regular screening would be reduced by 82%. In conclusion this risk-based selection criteria reduces mutation screening substantially while still enabling 61% of patients with mutations to be identified by regular mutation surveillance, and the remainder identified by evidence of emerging relapse manifested by a significant rise in the BCR-ABL level.

Author notes

Corresponding author