In chronic myeloid leukemia patients on second-line tyrosine kinase inhibitor therapy, deep sequencing of BCR-ABL1 at the time of warning may allow sensitive detection of emerging drug-resistant mutants

Abstract

Background: Imatinib-resistant chronic myeloid leukemia (CML) patients receiving second-line tyrosine kinase inhibitor (TKI) therapy with dasatinib or nilotinib have a higher risk of disease relapse and progression and not infrequently BCR-ABL1 kinase domain (KD) mutations are implicated in therapeutic failure. In this setting, earlier detection of emerging BCR-ABL1 KD mutations would offer greater chances of efficacy for subsequent salvage therapy and limit the biological consequences of full BCR-ABL1 kinase reactivation. Taking advantage of an already set up and validated next-generation deep amplicon sequencing (DS) assay, we aimed to assess whether DS may allow a larger window of detection of emerging BCR-ABL1 KD mutants predicting for an impending relapse.

Methods: a total of 125 longitudinal samples from 51 CML patients who had acquired dasatinib- or nilotinib-resistant mutations during second-line therapy were analyzed by DS from the time of failure and mutation detection by conventional sequencing backwards. BCR-ABL1/ABL1%(IS) transcript levels were used to define whether the patient had 'optimal response', 'warning' or 'failure' at the time of first mutation detection by DS.

Results: DS was able to backtrack dasatinib- or nilotinib-resistant mutations to the previous sample(s) in 23/51 (45 %) pts. Median mutation burden at the time of first detection by DS was 5.5 % (range, 1.5-17.5 %); median interval between detection by DS and detection by conventional sequencing was 3 months (range, 1-9 months). In 5 cases, the mutations were detectable at baseline. In the remaining cases, response level at the time mutations were first detected by DS could be defined as 'Warning' (according to the 2013 ELN definitions of response to 2nd-line therapy) in 13 cases, as 'Optimal response' in one case, as 'Failure' in 4 cases. No dasatinib- or nilotinib-resistant mutations were detected by DS in 15 randomly selected patients with 'warning' at various timepoints, that later turned into optimal responders with no treatment changes.

Conclusions: DS enables a larger window of detection of emerging BCR-ABL1 KD mutations predicting for an impending relapse. A 'Warning' response may represent a rational trigger, besides 'Failure', for DS-based mutation screening in CML patients undergoing second-line TKI therapy.

Keywords: BCR-ABL1; Chronic myeloid leukemia; Deep sequencing; Tyrosine kinase inhibitors; Warning.

Figures

Fig. 1

Backtracking dasatinib-resistant mutations by DS. Each line represents a patient and each circle corresponds to a sample. Full and empty circles indicate samples with mutations detectable or undetectable by DS, respectively. Light grey filling denotes samples in which the mutation was detectable by DS only. Dark grey filling denotes samples in which the mutation was detectable also by conventional sequencing. For each type of mutation, numbers in parentheses summarize the number of patients in which the mutation could be backtracked by DS/the total number of patients who acquired that type of mutation. Percentages indicate mutation relative abundance. ‘F’ means ‘Failure’, ‘W’ means ‘Warning’, ‘O’ means ‘Optimal’ response; ‘B’ means ‘Baseline’

Fig. 2

Backtracking nilotinib-resistant mutations by DS. See legend to Fig. 1 for explanations and abbreviations