Page 111 - 2022_01-Haematologica-web

RNA-sequencing of fusion genes in AML
the overlap of filtered fusion calls from Arriba and FusionCatcher. The built-in filter of Arriba excluded, on average, more putative false fusion events (74.8%) as com- pared to the built-in filter of FusionCatcher (62.3%). By applying our additional filtering strategies, we further reduced the amount of putative false fusion events substan- tially, resulting in an average of around 94% excluded fusion events from Arriba calls and around 96% from
A
FusionCatcher calls (Figure 2B). Besides detected true fusions (n=115), we also found 187 fusion events as robust candidates. Thirty of these 187 events have been described before, while 157 were putative novel fusion events (Online Supplementary Table S4). Clinical routine showed only low evidence for four of the 30 known events (Figure 1B, Online Supplementary Table S5), while 26 candidates were not reported by routine diagnostics in our cohorts. In two of the
B
C
Figure 1. Evidence for fusions by routine clinical diagnostics and RNA-sequencing. (A) True fusions detected by Karyotyping, molecular diagnostics (MDx) and RNA- sequencing (RNA-seq) in the AMLCG, DKTK, Beat AML and FIMM cohorts. Dark green boxes indicate high evidence, light green boxes indicate low evidence. Gray boxes represent no evidence although the respective method was performed. White boxes indicate that the respective method was not performed, or information was missing. (B) Known fusions detected with high evidence by RNA-seq that were missed or detected with low evidence only by Karyotyping/MDx. (C) Venn diagram summarizing fusions detected with the different methods.
haematologica | 2022; 107(1)
103