The genomic landscape, evolutionary trajectories and the mechanisms driving relapse in KMT2A-rearranged (KMT2A-r) infant and childhood acute lymphoblastic (ALL) and acute myeloid leukemia (AML) are not completely understood. We therefore studied 30 KMT2A-r ALL (n=16) and AML (n=14) patients of which 19 relapsed. Eleven diagnose-relapse-germline trios were analyzed by whole genome (WGS) and whole exome sequencing (WES), and all patients by ultra-deep resequencing (average 3400X) of patient-specific mutations identified by WGS/WES on the diagnostic (n=30) and relapse (n=11) samples, at diagnosis (n=30), during treatment (n=203) and at relapse (n=30, 1-4 bone marrow (BM) relapses in 17 patients).

The mutational burden increased from diagnosis to relapse in the diagnose-relapse-germline trios. The combined analysis of WGS/WES and ultra-deep data showed that the leukemias typically evolved through branching evolution, and that a selective sweep occurred at first relapse in 76% of patient. Further, additional subclones were often detected at relapse and in 69% the relapse clone was detected already at diagnosis as a subclone.

Eight pathways were recurrently altered including signalling, cell cycle, epigenetic, B-cell development, transcription factor, glucocorticoid receptor signalling, purine metabolism and cohesin. Signalling mutations were the most common type of mutations at diagnosis (50%) and at relapse (41%) and the diagnostic frequency was similar in patients that remained in complete remission (CR) and in those that relapsed (55% versus 47%). Signalling mutations were often subclonal (60%) and 30% had more than one mutation and mutations in signalling genes were usually gained or maintained at relapse in ALL but mainly lost or reduced in size in AML.

Alterations in cell cycle genes were enriched at relapse and either gained (60%) or maintained (27%), and included TP53 in ALL and AML, CDKN2A/B in ALL and CCND3 in AML. The TP53 alterations were typically biallelic at relapse. Only 2/30 patients had diagnostic TP53 mutations and both relapsed. Further, relapse-specific mutations in genes involved in glucocorticoid signalling and purine metabolism (CREBBP, NT5C2,PRPS2, NR3C1) were detected in all ALL patients with early relapse (n=4, during maintenance therapy). These relapse samples also had co-occurring IKZF1 deletions with 3 acquired at relapse and 1 maintained, as well as acquired TP53 mutations (3/4 patients). By contrast, infant ALL with a very early relapse (n=4, before maintenance therapy) lacked cell cycle and chemoresistance-associated mutations at relapse. Ultra-deep sequencing did not detect the CREBBP, NT5C2, PRPS2 or TP53 mutations at diagnosis and manual inspection of the WGS reads failed to detect the NR3C1 deletion. In AML, gain of WT1 mutations was seen in patients with late relapse (after >1 year in CR1).

Longitudinal sequencing on the 30 patients showed that at day 15, ALL patients that remained in CR (n=5) had higher levels of molecularly detectable leukemia cells as compared to those that relapsed (n=8), however at day 29, the reverse was seen with relapse patients having higher levels (n.s). Further, infants with very early relapse (n=3) had a higher leukemic burden at both time points as compared to those with early relapse (n=5). Patients with very early relapse typically reached CR later and the relapse clone could be detected at diagnosis in all 4 patients compared to 2/4 with early relapse. In 7 patients, 4 that relapsed and 3 that remained in CR, low-frequency mutations and KMT2A-fusion positive leukemic clones (DNA-level) were found at remission. Moreover, 4 patients had molecularly detectable clones across all investigated time points until relapse (6-13 time points), 2 had persistent disease but 2 entered CR. Finally, the relapse clone could be detected 46-126 days before relapse in 6 patients.

Combined, early relapse in infant and childhood KMT2A-r ALL was characterized by relapse-specific TP53, IKZF1, CREBBP, NR3C1, NTC52 or PRPS2 alterations. By contrast, very early relapse KMT2A-r infant ALL likely arises from pre-existing chemo resistant cells with a paucity of genetic alterations in relapse and chemoresistance-associated genes at relapse. Further, longitudinal molecular tracking provided biological insight into clonal response to treatment and captured the complexity of the temporal evolution in very early and early relapse patients.

No relevant conflicts of interest to declare.

Author notes

*

Asterisk with author names denotes non-ASH members.

Sign in via your Institution