In Chronic Myeloid Leukemia Patients on 2nd-Line Tyrosine Kinase Inhibitor Therapy, Deep Sequencing at the Time of Warning May Allow Sensitive Detection of Emerging BCR-ABL1 Mutants

Background and Aims: Next generation amplicon-based deep sequencing (DS) on the Roche, Illumina or Ion Torrent instruments is becoming accessible to a wider and wider number of diagnostic laboratories. Although conventional sequencing is still the gold standard, DS has been hailed by many as the future of diagnostic BCR-ABL1 kinase domain (KD) mutation screening. BCR-ABL1 KD mutations are infrequent in newly diagnosed chronic myeloid leukemia (CML) patients (pts) receiving 1st-line TKI therapy, but remain a challenge in relapsed pts, who usually display a greater genetic instability. Indeed, pts already harboring BCR-ABL1 KD mutations have a higher likelihood of developing additional, dasatinib (DAS)- or nilotinib (NIL)-resistant mutations – which is defined as a ‘failure’ by the 2013 European LeukemiaNet (ELN) recommendations. Taking advantage of a next-generation amplicon sequencing design and protocol set up and validated in the framework of the IRON-II international study, 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: among the imatinib (IM)-resistant CML pts who switched to 2nd-line TKI therapy and were referred to our laboratory for routine BCR-ABL1 transcript level monitoring and KD mutation screening by conventional sequencing, 51 acquired DAS- or NIL-resistant mutations after a median of 9 months (range, 3-27 months) of therapy and had leftover cDNA available at previous timepoints. To reconstruct the dynamics of mutation emergence, resequencing on a Roche GS Junior instrument was performed from the time of failure and mutation detection by conventional sequencing backwards. Runs were designed to achieve high sequencing depth, allowing reliable detection of variants down to 1% abundance. BCR-ABL1/ABL1%^IS transcript levels and/or cytogenetic response, whichever available, were used to define whether the patient had an ‘optimal response’, ‘warning’ or ‘failure’ at the time of first mutation detection by DS.

Results: baseline mutation status, as assessed by conventional sequencing, was available for all the 51 CML pts included in this retrospective study; 29/51 pts were positive for BCR-ABL1 KD mutations, with switch to NIL or DAS selected accordingly. Twenty-six pts were later found to have acquired DAS-resistant mutations (T315I, n=13; F317L/V, n=10; V299L, n=3) and 25 pts were later found to have acquired NIL-resistant mutations (T315I, n=4; F359V/I/C, n=7; Y253H, n=6; E255K, n=9; one patient acquired two mutations). DS was able to backtrack the DAS- or NIL-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% (range, 1-17%); median interval between detection by DS and detection by conventional sequencing was 3 months (range, 3-9 months). In 5 cases, the mutations were traceable at baseline; in the remaining cases, correlation with response at the time mutations were first detected by DS revealed a ‘warning’ according to the 2013 ELN definitions of response to 2nd-line therapy in 13 cases; an ‘optimal response’ in one case; a ‘failure’ in 4 cases. As a control, we used DS to explore BCR-ABL1 KD mutation status in 10 randomly selected pts with ‘warning’ at various timepoints, that later turned into optimal responses; no DAS- or NIL-resistant mutations were detected.

Conclusions: the 2011 ELN recommendations for mutation analysis suggest BCR-ABL1 KD to be screened by conventional sequencing in case of ‘failure’ of 2nd-line TKI therapy – according to the provisional definitions available at the time. Earlier detection of emerging BCR-ABL1 KD mutations allows a greater leeway in tackling drug resistance and enhancing therapeutic efficacy. Data presented herein indicate that:

1) DS may reliably pick TKI-resistant mutations earlier than conventional sequencing in a proportion of pts, and that

2) the recently introduced definitions of ‘warning’ may provide a rational trigger, besides ‘failure’, for DS-based BCR-ABL1 KD mutation screening in CML pts on 2nd-line TKI therapy.

A prospective cost-benefits evaluation of using DS in this and other settings is warranted, and will contribute useful information to the revision of the ELN recommendations for BCR-ABL1 KD mutation analysis.

Supported by: European LeukemiaNet, AIL, AIRC, FP7 NGS-PTL project, Progetto Regione-Università 2010-12 (L. Bolondi).