Backtracking BCR-ABL1 Mutants in Philadelphia-Positive Acute Lymphoblastic Leukemia Patients Relapsing on Tyrosine Kinase Inhibitors with Deep Sequencing: Implications for Routine Mutation Testing

Background: With the release of benchtop next-generation sequencers allowing to conjugate assay sensitivity and throughput with affordability, it was envisioned that NGS could soon replace conventional methods in the routine diagnostic identification of mutations relevant for diagnostication, prognostication and therapeutic tailoring. Having recently implemented and optimized a strategy for BCR-ABL1 kinase domain (KD) mutation screening based on the Roche-454 NGS technology, we decided to take advantage of its unique capability to capture the complexity and the clonal relationships of major and minor mutated populations present in a sample to focus on Philadelphia-positive (Ph+) acute lymphoblastic leukemias (ALL), where BCR-ABL1 KD mutations frequently arise and undermine the long-term efficacy of tyrosine kinase inhibitor (TKI)-based therapies.

Aims: Next generation amplicon-based deep sequencing (DS) was used i) to study the dynamics of expansion of BCR-ABL1 KD mutations in relation to BCR-ABL1 transcript levels in Ph+ ALL patients who relapse on TKIs and ii) to assess whether emerging mutations can be detected earlier than with conventional Sanger sequencing (SS).

Methods: This retrospective analysis was conducted on 35 Ph+ ALL patients who relapsed and acquired BCR-ABL1 KD mutations while receiving TKI-based therapy (imatinib, dasatinib, nilotinib) either first-line (Group A; n=10) or for recurrent disease (Group B; n=25). All the patients had been referred to our laboratory for minimal residual disease (MRD) follow-up monitoring by real-time reverse transcription-polymerase chain reaction and for BCR-ABL KD mutation analysis by SS in case of MRD positivity. To reconstruct the dynamics of mutation emergence, longitudinal re-analysis of bone marrow samples (n=113) from the time of hematologic relapse back to diagnosis (Group A) or back to TKI start (Group B) was performed on a Roche GS Junior instrument. Median time lapse between samples was 6 weeks (range, 4-15 weeks). Runs were designed to achieve high sequencing depth, allowing reliable detection of sequence variants at least down to 1% abundance.

Results: DS was found to provide a more accurate picture of BCR-ABL1 KD mutation status, both in terms of presence/absence of mutations and in terms of clonal complexity. DS indeed detected low level mutations, alone or in addition to SS-detectable mutations, in 14/35 samples from patients in Group A and 24/72 samples from patients in Group B. In all the cases with two or more mutations, DS uncovered polyclonal mixtures of single and compound mutants, with up to five distinct mutant subclones detectable. With very few exceptions, compound mutants were found to represent the dominant clone. Patients in Group B, who had received two or even three lines of TKI therapy, generally displayed greater complexity than patients in Group A.

TKI-resistant mutants were found to undergo rapid selection in Ph+ ALL patients. In 13/35 (37%) patients, however, DS would have allowed to detect emerging mutations earlier than SS actually did and before hematologic relapse – including a Y253H mutation in a sample collected at diagnosis from a patient who relapsed eight weeks later.

Detection of emerging TKI-resistant mutations by DS was accompanied by total BCR-ABL1/ABL1 transcript increase from half a logarithm to two logarithms, or preceded by a shift from undetectable to detectable BCR-ABL1 transcripts. Even low MRD levels were found to hide low level TKI-resistant mutants.

Conclusions: The enhanced sensitivity as well as the possibility to identify compound mutations point to next generation amplicon-based deep sequencing as an ideal alternative to conventional sequencing for BCR-ABL1 KD mutation screening of Ph+ ALL patients receiving TKI based therapies. Cost is not a concern even for small/medium scale laboratories, considering that many samples and many genes can be pooled and analyzed simultaneously. A turnaround time of 3-4 days is generally feasible, irrespective of chemistries and instruments, allowing DS results to be practically integrated in the clinical decision algorithm. Further, prospective evaluation of DS for BCR-ABL1 KD sequence surveillance in the framework of a clinical trial of TKI therapy and MRD monitoring of Ph+ ALL patients is warranted.

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