Objectives Interrogate the dynamic evolution of cancer-related gene mutations in subjects with chronic myeloid leukaemia (CML) receiving the third generation- (3G-) tyrosine kinase inhibitor (TKI) therapy, and define those mediating resistance. Methods 176 consecutive subjects with chronic or accelerated phase CML failing imatinib and/or the second generation- (2G-) TKI and receiving a 3G-TKI (ponatinib or olverembatinib) were studied. Deep targeted sequencing was used to screen cancer-related gene mutations and Sanger sequencing for ABL mutations. Results 176 subjects in chronic (n = 136) or accelerated (n = 41) phases received ponatinib (n = 147) or olverembatinib (n = 29) therapy. 19 subjects were exposed to imatinib priorly; 16, the 2G-TKI; 141, imatinib and the 2G-TKI. 142 (81%) had ABL mutations including ABLT315I (n = 87, 61%), ABLT315I+additional (n = 30, 21%) or ABLothers (n = 25, 18%) at baseline. 157 (89%) subjects had other cancer-related gene mutations including ASXL1 (n = 121, 69%), KMT2D (n = 14, 8%), RUNX1 (n = 12, 7%), PHF6 (n = 9, 5%), KMT2C (n = 8, 5%), IKZF1 (n = 8, 5%), STAT5A (n = 8, 5%), TET2 (n = 8, 5%), DNMT3A (n = 7, 4%) and KDM6A (n = 7, 4%) and so on at baseline. In multivariable analyses, ASXL1G646Wfs*12 and PHF6 mutations were significantly-associated with the poor cytogenetic and molecular responses; RUNX1 and PHF6 mutations with poor progression-free survival; IKZF1 and STAT5A mutations with poor survival. ABLT315I mutation was significantly-associated with good responses. Follow-up analyses were done in 105 subjects. According to the therapy response of 3G-TKI, 51 subjects were classified into responsive cohort; 54 subjects, resistant cohort including primary (n = 44) and secondary (n = 10) resistance. In the responsive cohort, only 1 subject retained their ABL mutation and none developed a new ABL mutation (Figure 1B). In the resistant cohort, 6 subjects developed new ABL mutations including E255K/V (n = 2), F359V/I (n = 2), G250E and T315M (Figure 1C). There was a significantly-higher proportion of subjects with ABL mutations in the resistant cohort compared with the responsive cohort (52% vs. 2%, p < 0.001) despite similar ABL mutation status at baseline. As for other cancer-related gene mutations, 92 (88%) had other cancer-related gene mutations with a median number of 2 (range, 0 - 7) at baseline. Compared with the responsive cohort, the detected frequency and the median number of mutations in the resistant cohort were significantly-higher both at the baseline (p < 0.001 - 0.027) and at the end of observation period (p < 0.001 - 0.002) (Figure 2&3). More subjects developed new mutations in the resistant cohort than those in the responsive cohort (74% vs. 39%, p < 0.001, Figure 2&3). As the most frequently detected of ASXL1 mutation, despite similar frequency of ASXL1 mutation between the responsive and resistant cohorts at baseline (p = 0.68), more subjects in the resistant cohort had ASXL1G646Wfs*12 mutation than those in the responsive cohort (p = 0.03), while similar frequency of ASXL1non-G646Wfs*12 mutations. At the end of observation, 52 (50%) subjects (including 40 in the resistant cohort and 12 in the responsive cohort) had the ASXL1 mutation. In subjects with ASXL1 mutation, 21 subjects were detected to the ASXL1G646Wfs*12 mutation, with only 1 (2%) subject in the responsive cohort, 20 (37%) in the resistant cohort (p < 0.001). Frequencies of ASXL1non-G646Wfs*12 were similar. In addition, the variant allele frequencies (VAFs) of ASXL1G646Wfs*12 and ASXL1non-G646Wfs*12 mutations were significantly-decreased during 3G-TKI therapy in the responsive cohort (p < 0.001 - 0.002, Figure 4) but not in the resistant cohort. There were no significant differences in the primary resistance cohort during the 3G-TKI therapy. However, the VAFs of ASXL1G646Wfs*12 mutation significantly-decreased when in the temporarily response (p = 0.05), while significantly-increased when the secondary resistance occurred (p = 0.015, Figure 4A) in the secondary resistant cohort. In contrast, the VAFs of ASXL1non-G646Wfs*12 mutation presented the decreased trend when in both temporarily response and secondary resistance (Figure 4B). The patterns of the dynamic evolution in RUNX1, IKZF1, SETBP1 and PHF6 mutations were similar as those in ASXL1 G646Wfs*12 mutation. ConclusionsASXL1G646Wfs*12, RUNX1, IKZF1, SETBP1 and PHF6 mutations are the key drivers of resistance to 3G-TKI therapy in CML. Figure 1View largeDownload PPTFigure 1View largeDownload PPT Close modal
Read full abstract