Paper Details

ABSTRACT

Background
Outcomes for patients (pts) in blast phase (BP) CML remain poor. Genomic factors are involved in BP progression, with variants in ASXL1, RUNX1 and IKZF1 being the most common. ASXL1 variants (ASXL1+) in BP are linked to reduced survival. Somatic variants have been reported in almost all pts in BP, with a significant portion consisting of additional gene fusions and deletions. The molecular heterogeneity observed in BP presents a complex therapeutic challenge. Combining many genomic datasets derived from different centers through the iCMLf Genomics Alliance could facilitate greater understanding of the molecular events to improve treatment strategies.

Methods
The Harmony Platform is a central repository of data from multiple contributors. Whole exome or targeted gene panel data of pts in BP from 7 centers was submitted: Europe 3, Taiwan, Canada, Kuwait and Australia. Variants in non-BCR::ABL1 genes meeting strict pathogenicity criteria were included, herein termed variants. Methods were capable of detecting single nucleotide variants, including BCR::ABL1 mutations (ABL1+), and small insertions/deletions, but not gene fusions or large deletions. Frequencies were compared using the Chi-square or Fisher’s exact tests. Kaplan–Meier curves estimated overall survival (OS), and the log-rank test was used to assess differences in survival distributions between groups.

Results
The dataset analyzed comprised 120 pts with BP progression after diagnosis in chronic (n=111) or accelerated phase (n=9). BP lineage was available for 88% of pts: myeloid BP (MBP) 63% and lymphoid BP (LBP) 37%. Frontline TKI was received by 68/120 pts (57%) and 79% of these pts received imatinib. Of the 52 pts not treated with frontline TKI, 92% were diagnosed in the pre-TKI era and most received interferon.

The variant yield in BP was 64%, which was lower than previously reported, presumably due to the inability to detect gene fusions and deletions. ASXL1+ and RUNX1+ dominated BP: 24% and 20% of pts. Collectively, these 2 genes were mutated in 58% of pts with detectable variants. In the remaining pts with variants, 29 other genes were mutated. ABL1+ was detected in 49% of evaluable pts and 71% of these pts had cooccurring non-BCR::ABL1 variants. For 91 pts who had matched diagnosis and BP data, the variant yield increased from diagnosis to BP: 34% to 64%, P=0.0001, as did the number of pts with multiple variants: 11% to 31%, P=0.001.

The genomic profile in BP was assessed based on frontline TKI therapy and BP lineage. There was no difference in the overall variant rate in BP if treated with frontline TKI. However, ASXL1+ was significantly lower in pts receiving frontline TKI, 15% vs 37%, P=0.006. In contrast, there was no difference for RUNX1+: 21% vs 19%. ASXL1+ also occurred less often in LBP than MBP: 5.6% vs 35%, P=0.0007. For pts in MBP, those receiving frontline TKI had notably lower frequency of ASXL1+, although not reaching significance: 24% vs 41%, P=0.19. Frontline TKI did not influence the frequency of RUNX1+ for pts in MBP: 20% vs 21%, P=0.97.

Interestingly, the rate of MBP was lower for pts receiving frontline TKI than those without frontline TKI: 47% vs 85%, P=0.0001. Conversely, the rate of LBP was higher in pts treated with frontline TKI: 53% vs 12%, P=0.0001, suggesting frontline TKI could modify BP lineage frequency.

The 12 month OS probability was 32% after BP progression. Subgroup analysis revealed differences in OS, although not reaching significance in all cases, which may be linked to low sample number. Among pts treated with frontline TKI, 12 month OS was higher with frontline TKI vs no frontline TKI, 41% vs 24%, P=.076. 12 month OS was higher in LBP vs MPB: 44% vs 26%, P=0.11. Significantly, the 12 month OS was lower for pts with ASXL1+ vs ASXL1-: 19% vs 37%, P=0.04. In the subgroup of pts in MBP, the OS was also lower for patients with ASXL1+: 18% vs 30%, but not for RUNX1+: 23% vs 26%. No other factor showed an apparent impact on OS.

Conclusion
The study represents the first genomic analysis to investigate the effect of frontline TKI therapy on the variant profile in BP. Pts with ASXL1+ in BP had lower 12 month OS, and frontline TKI was linked with a lower frequency of ASXL1+. Intriguingly, frontline TKI modified the frequency of MBP. This research prompts an inquiry into whether the suppression of ASXL1+ with frontline TKI therapy could contribute to lower rates of BP transformation.