Targeted RNA Sequencing Assay Efficiently Identifies Cryptic KMT2A (MLL)-Fusions in Acute Leukemia Patients

The KMT2A gene is rearranged in pediatric and adult acute lymphoblastic leukemia (ALL) and acute myeloid leukemia (AML). By fusing with a number of different partner genes, the KMT2A gene generates chimeric proteins with oncogenic abilities. Due to the promiscuity of the gene, routine genetic analysis of KMT2A-rearranged leukemia is time-consuming as it often involves chromosome and FISH analyses, RT-PCR or southern blot. We have thus developed a captured-based next generation sequencing approach to detect KMT2A fusion genes, especially useful when routine methods indicate a rearrangement but cannot identify the specific fusion partner.

First, we analyzed bone marrow samples from two AML cases with deletions of the distal part of the KMT2A gene and, as a control, a pre B-ALL with a known KMT2A-AFF1 fusion. The capture was carried out using custom hybridization probes (SureSelect, Agilent) encompassing the entire KMT2A gene. Sequencing was performed on the Ion Torrent PGM, a sequencing instrument well suited for routine diagnostics given its rapid turn-around time. Putative fusions were identified from sequencing reads connecting an exon within KMT2A to an exon from any other gene and the results were confirmed by RT-PCR. As comparison we performed the standard RNA-sequencing protocol using Ion Proton. In the two patients with unknown KMT2A-rearrangements, the capture approach identified 3034 and 2118 reads over a KMT2A-MLLT4 gene fusion, while only 4 and 2 reads, were detected with the standard RNA-sequencing protocol (Table 1). Similarly, RNA-sequencing and the capture approach detected in the control a KMT2A-AFF1 fusion, with a sequencing depth of 4 and 144 reads, respectively. Interestingly, with the capture approach we obtained 5032 reads encompassing the reciprocal fusion, AFF1-KMT2A, underlining the importance of the reciprocal transcript in this entity. Hence, the KMT2A capture approach resulted in a 5000-35000-fold enrichment of the KMT2A fusions compared to standard RNA-sequencing. Although the standard RNA-sequencing detected reads covering the fusion genes, it yielded far too low sequencing depth to be of diagnostic use.

We then performed the KMT2A capture approach on three additional cases with distal KMT2A deletions. In one case, we detected a KMT2A-ARHGEF12 fusion, resulting from a 20Mb interstitial deletion (Table 1). This aberration is missed by standard cytogenetics and consequently likely under-diagnosed. However, in two other cases no KMT2A-fusions were identified. In these cases, the 11q-deletion was present only in a proportion of cells and in one, the deletion was absent at relapse, indicating that it was not likely the primary event in this leukemia. These results demonstrate the importance of establishing if a suspected KMT2A-rearrangement truly generates a KMT2A-fusion.

In summary, the hybridization-based capture sequencing approach efficiently detects KMT2A-fusion transcripts and reliably identifies canonical and cryptic KMT2A-rearrangements, generating significantly higher number of fusion breakpoint reads, as compared to RNA-sequencing. Furthermore, sensitive RT-PCRs for follow-up (MRD) can be readily designed using the sequence information.

All results are from bone marrow samples taken at time of diagnosis.^aKMT2A specific FISH-analysis was performed using Vysis KMT2A break apart probe. ^bRNA-sequencing not performed. ^c No KMT2A-fusion detected.