Prospective Next-Generation Sequencing of Mutations in the ABL1 Kinase Domain in CML Patients Lacking Optimal Response According to ELN Guidelines

Background: In chronic myeloid leukemia (CML), mutations in the ABL1 kinase domain cause resistance to tyrosine kinase inhibitors (TKIs), and require therapy intervention. According to ELN guidelines (Baccarani et al., 2013), ABL1 sequencing is recommended for cases of progression, failure and warning. Failure and warning apply to patients, who did not reach treatment milestones or had shown response and subsequently lost it. An increase in BCR-ABL1/ABL1 ratio can be regarded as first alert. Failure is then defined as loss of major molecular response (BCR-ABL1/ABL1 <0.1%) if confirmed in two consecutive tests (one with BCR-ABL1/ABL1 >1%). The recommendations aim at Sanger sequencing (sensitivity limit of 15% mutated BCR-ABL1 transcripts) and claim more data collection on the use of highly sensitive next-generation sequencing (NGS).

Aim: We prospectively performed NGS to evaluate the frequency of mutations and the need to sequence patients within the failure category, the warning zone and patients, who did not yet meet warning or failure criteria.

Patients and methods: We selected 83 CML patients with a p210 transcript (median age: 68; range: 17-87; 38 females; 45 males). Patients were under TKI treatment at the analyzed time point and fulfilled at least one of the following criteria: a) tenfold increase in BCR-ABL1/ABL1 ratio under any line of treatment, or b) not reaching BCR-ABL1/ABL1 ratio of 1% after 6 or 0.1% after 12 months. Patients had undergone first-line TKI treatment (n=59), or first received imatinib and thereafter a second line TKI (dasatinib or nilotinib, n=13) or multiple lines of treatment including ponatinib and bosutinib (n=11). To compare sequencing methods, we used 14 BCR-ABL1 positive cDNAs, for which we had identified mutations by Sanger sequencing. Next, NGS was performed on the test set and the entire CML cohort on the MiSeq platform (Illumina, San Diego, CA): 6 amplicons covered amino acid 184-510 and were generated from preamplified BCR-ABL1 (minimal sequencing depth: 250X)

Results: As technical comparison, 20 mutations (in 14 cDNAs), previously detected by Sanger sequencing, were analyzed by NGS. We detected all mutations and correlation of mutation burdens was high (linear regression: R²=0.975; P<0.01). By sequencing triplicates of 12 known mutations, we could show that 3% of mutated fusion transcripts can be detected by NGS (with BCR-ABL1 >1%).

Using NGS, 11/83 patients of our CML cohort showed resistance mutations, of which T315I was the most frequent aberration (n=5). Two patients had two mutations (both T315I+Y253H). One patient had only a 6% mutation burden, which was not detected by Sanger sequencing. In the other ten patients, mutations could have been detected by Sanger sequencing, too (median: 98%; range 6%-100%). Due to our cohort selection criteria, we sequenced 23 patients with a BCR-ABL1/ABL1 ratio between 0.1% and 1%. Here, mutations were already detectable in three patients (13%). For comparison: 8/60 (13%) were mutated in the cohort with BCR-ABL1>1%.

We did not detect any mutation in 18 patients without optimal response after 6 months. In 25 patients, lacking optimal response after one year (warning, n=13; failure n=12), we identified two (8%) mutated cases (both warning zone). Of patients with more than one year of treatment, 2/12 (17%) with TKI failure, and 1/6 (17%) from the warning zone were mutated. In contrast, we detected 5/11 (45%) mutated cases in the subgroup of patients who had undergone a tenfold increase in BCR-ABL1 transcripts despite multiple lines of TKI treatment.

For 11 patients without a mutation, we received a follow-up sample (median interval between samples: 9 weeks). None of the patients had gained a novel mutation. This underlines that our initial sequencing approach by NGS did not leave mutations undetected, which would have required a longer CML cell growth for detection. Therefore, we sequenced patients with a tenfold increase in BCR-ABL1/ABL1 ratio before a second test was conducted (not yet in failure category) and found already one (9%) mutated case.

Conclusions: 1) Mutations in the ABL1 kinase domain can be detected earlier and by higher sensitivity using NGS. 2) This has clinical implications and should lead to mutation screening whenever a strong increase of BCR-ABL1 transcripts is seen and in the warning zone (according to ELN guidelines). 3) TKI trials should study Sanger sequencing and NGS in parallel to define best approaches.