Detection and Monitoring of BCR-ABL1 Kinase Domain Mutations By Next Generation Sequencing

Background.  Among myeloproliferative diseases, development of chronic myeloid leukaemia (CML) is associated with the emergence of the fusion oncogene BCR-ABL1 resulting from a t(9,22) chromosomal translocation (Philadelphia chromosome). Mutations of the BCR-ABL1 kinase domain constitute a major cause of treatment failure in CML patients receiving tyrosine kinase inhibitor (TKI) treatment. Moreover, the occurrence of cells with multiple mutations is frequently associated with a higher resistance rate to the different TKI (Imatinib, Dasatinib or Nilotinib). So far, the gold standard procedure to detect BCR-ABL1 mutations remains the conventional Sanger Sequencing (SS), endowed with an analytical sensitivity of 10 to 20 %. The recent implementation of Next Generation Sequencing (NGS) allows lowering the sensitivity level and quantitative follow-up of the mutated subclone(s), which probably improves CML patient’s treatments management.

                Aims.  In this retrospective study, we evaluate the advantage of NGS approach to i) identify patients harbouring (low level) mutations that have not been not assessed by conventional methods, ii)  detect the emergence of mutated clones earlier than SS and iii) monitor evolution of mutations.

                Methods.  Total BCR-ABL1 RNA was transcribed into a long range cDNA covering the kinase, the regulatory, and the SH2/SH3 domains of either p190 or p210 BCR-ABL1 transcripts (exons 4 to 10). From primers designed with the Ampliseq^TM Designer Software, a set of 10 amplicons was generated according to the AmpliSeq^TM protocol. Bar-coded libraries were sequenced on the Ion Torrent PGM platform and data were analysed with Torrent Suite and NextGene softwares. Serial dilutions of samples harbouring mutations at different levels were used to determine a variant frequency cut-off. Our methodology was applied to a group of 36 patients presenting with poor response to TKI and with no mutation detected by SS and to a set of 100 samples, corresponding to 20 mutated patients, at different time points before the time of mutation identification by SS.

                Results.  From the serial dilutions experiment, the detection limit of the assay was set up to 2 % (R² > 0.997). An overall coverage ranging from 20 000 to 50 000 reads can be achieved for the hotspot mutations when up to 12 samples were tested together on a 316 Ion chip.  On the 36 patients tested by NGS versus SS, no one was found to harbour TKI-resistance mutation.  NGS successfully detected all mutations identified by SS; mutations were typically detected within 4 months (18/20 patients) and were also detected up to 9 months prior to detection by SS, even in patients with a low abundance of BCR-ABL1 transcripts and in sequencing failure by SS.   In 2 patients presenting with up to 3 mutations, evolution of mutations (emergence, expansion or depletion) correlates with clinical data of treatment decisions, i.e.  E255K (patient-1) and L248V (patient-2) depletions when switching from Imatinib to Dasatinib,  F317L (patient-1), G250E (patient-1) and T315I (patient-2) expansions under Dasatinib and a complete but transient depletion of T315I (patient-2) with the protein synthesis inhibitor homoharringtonine (Omacetaxine).  Finally, assessment of the mutation status of one patient with compound mutations following an Illumina protocol on a MiSeq platform had allowed comparison of technologies performances.

                Conclusions.  NGS did not detect mutations in 36 patients poorly responding to TKI with no detectable BCR-ABL1 TK mutation by SS. For 20 patients showing BCR-ABL1 TK mutation by SS, NGS was able to detect the mutation in samples taken up to 9 months prior to the moment when the mutation was observed by SS. Advances in sequencing technologies and further lowering sensitivity levels can contribute even more to earlier detection of mutations and guide an earlier switch of TKI. Quantitative and sensitive monitoring of mutation evolution can also inform the most appropriate and optimized treatment algorithms. A prospective evaluation of the clinical impact of NGS-based BCR-ABL1 mutation detection is ongoing.