Phasing Analysis Of TKI Resistance Mutations In The BCR-ABL1 Kinase Domain and Neighboring Domains Using Next-Generation Sequencing

Many CML patients treated with tyrosine kinase inhibitors (TKIs) eventually develop resistance as a result of ABL1 kinase domain (KD) mutations, and sequential treatment with different TKIs may select for multiple BCR-ABL1 mutations. Whether multiple mutations arise in distinct clones (in trans, or polyclonal mutations) or instead are present within the same BCR-ABL1 molecule (in cis, or compound mutations), has been shown to have important implications with respect to TKI sensitivities (Eide, C.A. et al., Blood 2011). Distinguishing between polyclonal and compound mutations, or mutation phasing, for the ABL1 KD has not been clinically practical with standard mutation detection methods. Here we have developed a highly sensitive next-generation sequencing (NGS) assay on the Ion Torrent PGM along with a proprietary data analysis pipeline that together enable deep sequencing of the BCR-ABL1 KD and neighboring domains with a 1% limit of detection and quantitative reporting of mutation phasing.

RT and long range PCR was performed to amplify BCR-ABL1 e1a2/3, e13a2/3, and e14a2/3 fusion transcripts and the PCR products were enzymatically randomly fragmented and ligated with Ion Torrent sequencing adaptors. Size-selected libraries were quantified, pooled, amplified with the OneTouch system and sequenced with the Ion Torrent PGM using 400 bp sequencing chemistry. Sequencing data were analyzed with Torrent Suite 3.4.2 with variant frequency cutoff adjusted to 1%. Variants were further annotated with a proprietary analysis pipeline and variant report was produced after manually reviewing variants by Integrative Genomics Viewer. If more than one non-synonymous variant was reported in a sample, a proprietary phasing analysis pipeline was applied to report the mutation spectrum of all of the combinations of multiple mutations in the sample.

To validate the accuracy of the sequencing method which employs 400 bp sequencing chemistry, we compared this assay with our previously validated BCR-ABL1 NGS assay based on Ion Torrent 200 bp sequencing chemistry for a set of clinical specimens from CML patients previously treated with TKI. Results were highly concordant and similarly sensitive, with 11/11 variants (frequencies ranging from 2% to 100%) identified with comparable frequencies by both methods.

To evaluate the specificity of the phasing analysis, an artificial sample was created by mixing two samples (b2_10 and b2_6) with 6 distinct variants present at ratio of 1:19. Because variants are unique in each sample, any compound mutation composed of b2_10 variant and b2_6 variant identified would be false positive. The false positive error rate (percentage of b2_6 variant as compound mutation with b2_10 variant and vice versa) ranged from 0 – 0.6%, which was consistent with sequencing error rate. We conservatively define a compound mutation as true if it is present in at least 5% of any one of the component variants in the compound mutation. Mutation detection and phasing analysis were reproducible on different chips (314 v2 and 318 v2) and different library preps from the same long range PCR product of BCR-ABL1.

Table 1 shows the mutation spectrum from sample b2_6. Of the four variants detected, L248V and G250E were mutually exclusive (in trans), while T315I and M351T were present as compound mutations with each other and, separately, with either L248V or G250E. Notably, >86% of the molecules harbored single mutations, and no compound mutations containing more than 2 variants were observed.

We have developed and validated a sensitive NGS assay that enables deep sequencing of the BCR-ABL1 KD and neighboring domains along with quantitative mutational phasing. This method has been applied in evaluating >250 clinical specimens for a clinical trial of a third-generation TKI(results reported separately). The ability to easily determine the mutation phasing of a CML patients’ mutation profile using this assay will allow for investigations into compound mutation-based resistance mechanisms and may be used to better guide treatment decisions.

Li:MolecularMD: Employment. Yan:MolecularMD: Employment. Darwanto:MolecularMD: Employment. Fang:MolecularMD: Employment. Liu:MolecularMD: Employment. Drafahl:MolecularMD: Employment. Toplin:MolecularMD: Employment. Spittle:MolecularMD: Employment. Galderisi:MolecularMD: Employment.