Next Generation Sequencing-Based BCR-ABL1 Kinase Domain Mutation Screening in De Novo and Tyrosine Kinase Inhibitor-Resistant Philadelphia Chromosome-Positive Acute Lymphoblastic Leukemia: Results of a Prospective Study

In Philadelphia-positive (Ph+) Acute Lymphoblastic Leukemia (ALL) patients (pts), resistance to tyrosine kinase inhibitors (TKIs) is frequently associated with the selection of one or more mutations in the BCR-ABL1 kinase domain (KD). The swift emergence of mutant clones as early as during induction therapy supports the hypothesis that, at least in some cases, mutations may already be present at diagnosis. Next Generaton Sequencing (NGS) has been proposed as an alternative to Sanger sequencing (seq) for BCR-ABL1 KD mutation screening because of its greater sensitivity and accuracy, but no studies have so far evaluated its prospective use in Ph+ ALL.

Between 2015 and 2018, we have used NGS in parallel to Sanger seq to analyze a consecutive series of 126 Ph+ ALL pts who were newly diagnosed (n=39) or who had relapsed/refractory disease (n=87) on TKI therapy. In 22 cases, both bone marrow and peripheral blood were analyzed and compared. NGS of ≈400bp amplicons generated by nested RT-PCR was performed on a Roche GS Junior (until April 2017) or on an Illumina MiSeq (from May 2017 on). Read alignment and variant calling (with a lower limit set to 3%) were done with the AmpSuite software (SmartSeq srl). When multiple mutations mapped within the same sequence reads, assessment of cis vs trans configuration was done correcting for the probability of PCR recombination.

Three out of 39 (7.7%) de novo Ph+ ALL pts had low burden point mutations detectable by NGS: one had a V289A (variant frequency, 3.4%); one had a D276G (4.0%) and a F359V (3.5%); one had an E255K mutation (3.3%). The first pt was enrolled in the GIMEMA LAL1811 study of frontline ponatinib; the second and the third pts were enrolled in the GIMEMA D-ALBA study of frontline sequential treatment with dasatinib and blinatumomab. All pts achieved molecular remission, consistently with the mutations being sensitive to the TKIs received.

The 35INS insertion/truncation mutant was detected in 27 (69%) pts, who all have so far achieved molecular remission. This is in line with the report by O'Hare et al (Blood 2011) suggesting that the 35INS variant is kinase-inactive and does not contribute to TKI resistance. For this reason, the 35INS was excluded from subsequent analyses.

Relapsed/refractory pts positive for mutations by Sanger seq were 57 (65%); those positive for mutations by NGS were 69 (79%). Fifty-six out of 87 (49%) pts had >1 mutation (up to 13) detected by NGS. NGS identified low burden mutations (i.e., mutations present in a proportion of transcripts between 3 and 20%) in 12 pts who were negative for mutations by Sanger seq. Most importantly, NGS provided a more accurate picture of BCR-ABL1 mutations status in 40 (46%) pts who turned out to have one or more low burden mutations in addition to the dominant mutation(s) detectable by Sanger seq. In all cases, each low burden mutation detected by NGS could be recognized as poorly sensitive either to the TKI the pt was receiving at the time of testing, or to the previous TKI. The clonal nature of NGS-based analysis further proved its utility i) in 4 pts where Sanger seq had shown 2 base substitutions in the same codon so that the actual amino-acid change(s) were impossible to infer (a ponatinib-resistant pt with a T315M mutation, 2 dasatinib-resistant pts with various combinations of F317I, F317C and/or F1317L, a dasatinib-resistant pt with 2 different nucleotide substitutions both leading to the V299L), and ii) in 48/56 pts who had ≥2 mutations whose clonal configuration could not be resolved. Twenty-eight out of these 48 pts were found to carry one or more (up to 3) compound mutants. Compound mutants were more common in pts who had failed ≥2 lines of therapy, whereas polyclonality was more common in pts who had failed first line therapy. The most frequent compound mutants were T315I+E255K and T315I+E255V. Interestingly, the latter was associated with poor or no response to ponatinib.

Our results in a relatively large series of Ph+ ALL pts suggest that an NGS-based approach provides a more accurate characterization of the complexity of BCR-ABL1 KD mutation status, including compound mutants some of whom may be poorly sensitive even to ponatinib. Mutations may already be detected at the time of diagnosis. It remains to be assessed whether more sensitive techniques like digital PCR may identify a greater number of pts with pre-therapy mutations and whether the detection of pre-therapy mutations may be used to guide 1st-line treatment selection.