BCR-ABL1 compound mutations in tyrosine kinase inhibitor-resistant CML: frequency and clonal relationships

Abstract

BCR-ABL1 compound mutations can confer high-level resistance to imatinib and other ABL1 tyrosine kinase inhibitors (TKIs). The third-generation ABL1 TKI ponatinib is effective against BCR-ABL1 point mutants individually, but remains vulnerable to certain BCR-ABL1 compound mutants. To determine the frequency of compound mutations among chronic myeloid leukemia patients on ABL1 TKI therapy, in the present study, we examined a collection of patient samples (N = 47) with clear evidence of 2 BCR-ABL1 kinase domain mutations by direct sequencing. Using a cloning and sequencing method, we found that 70% (33/47) of double mutations detected by direct sequencing were compound mutations. Sequential, branching, and parallel routes to compound mutations were common. In addition, our approach revealed individual and compound mutations not detectable by direct sequencing. The frequency of clones harboring compound mutations with more than 2 missense mutations was low (10%), whereas the likelihood of silent mutations increased disproportionately with the total number of mutations per clone, suggesting a limited tolerance for BCR-ABL1 kinase domain missense mutations. We conclude that compound mutations are common in patients with sequencing evidence for 2 BCR-ABL1 mutations and frequently reflect a highly complex clonal network, the evolution of which may be limited by the negative impact of missense mutations on kinase function.

Figure 1

Polyclonal versus compound mutations. In a subset of patients who develop clinical resistance to ABL1 TKIs, more than 1 point mutation in the kinase domain of BCR-ABL1 is detectable by direct sequencing. In the case of polyclonal mutations, these BCR-ABL1 mutations (green and red stars; top panel) exist separately in different clones. In contrast, BCR-ABL1 compound mutants exhibit 2 mutations within the same BCR-ABL1 molecule (green and red stars; bottom panel).

Figure 2

Mutational patterns revealed by cloning and sequencing of the BCR-ABL1 kinase domain region. Hypothetical mutation identities are designated A through H for discussion/explanation purposes. The range of mutation patterns revealed by cloning and sequencing when 2 BCR-ABL1 kinase domain mutations, A and B (green and red stars in the trace), are detected by direct sequencing are represented. This model encompasses all observed mutational patterns among the patients with 2 mutations evident by direct sequencing in this study. Detection of mutations A and B by direct sequencing can reflect their presence in the same BCR-ABL1 molecule (detected in 1 clone) or in different BCR-ABL1 molecules (detected in separate clones). The presence of both A and B in the same or different clones is signified as A/B or as A and B, respectively. In some compound mutant patients, clone A and clone B were observed separately in addition to clone A/B. Mutation A or B was sometimes coexistent with add-on mutations, for example, A/D and B/E. Predominant compound mutants could also acquire further add-on mutations as exemplified by A/B/C. Some mutations, such as F (single independent low-level mutants) or G/H (compound independent low-level mutants), were observed in clones that did not carry either predominant mutation, A or B.

Figure 3

Successive acquisition of mutations. Five serial samples were available for patient CML#19. The clones with thick circles represent the mutations detected by direct sequencing. The number of sequenced clones for each sample is shown on the left. The number of each mutated clone is shown inside each clone unless it was detected only once. The unmutated or independent low-level mutant clones are not shown here. M351T and Y253F were detected by direct sequencing in the first sample under imatinib (IM) therapy (41 months). Cloning and sequencing confirmed M351T/Y253F as a compound mutation but also revealed other low-level clones presumably derived from M351T that were undetectable by direct sequencing. The red line indicates a 30-month gap between stoppage of imatinib therapy and start of nilotinib (NI) therapy, during which time the patient was treated with homoharringtonine. Direct sequencing in the next 3 samples taken after the start of nilotinib therapy detected only M351T. Similar to the first sample, cloning and sequencing revealed add-on compound mutants derived from M351T not detected by direct sequencing. An E255K/M351T compound mutation was first observed under dasatinib (DA) therapy. E255K/M351T was the only mutation detected in the last sample, and was evident by direct sequencing and the more sensitive cloning and sequencing method. M351T/E255K developed along with resistance to dasatinib in this patient. The patient's disease progressed to blast crisis a few months after detection of M351T/E255K and loss of response to dasatinib.

Figure 4

Simultaneous evolution of compound mutant clones at comparable levels. Cloning and sequencing of 55 colonies for CML#14 revealed a complex pattern of compound mutations. There were at least 3 major classes of clones (F359V/T315I, K294K/M351T/V299L, and E355G) and their derivatives, which formed the majority of the BCR-ABL1 clones in this sample. The thick red, green, and blue circles represent 11, 6, and 14 clones, respectively. The dotted circle represents a clone that was not detected but was expected to be the parent of the 2 observed clones. The remaining circles represent clones detected a single time. These single clone results should be interpreted with caution because PCR error can never be rigorously excluded. Silent mutations are labeled in red. There is a marked increase in the proportion of silent mutations as clones acquire additional mutations. One clone presumably destined for immediate elimination has a stop mutation (E355G/Q300Stop) expected to abolish kinase activity in the middle of the kinase domain.

Figure 5

The percentage of silent mutations increases disproportionately with the total number of mutations per clone. The x-axis represents the number of mutations per clone and the y-axis represents the percentage of clones with at least 1 silent mutation (gray bars) compared with the expected percentage (white bars).