While the outcome for children with acute lymphoblastic leukemia (ALL) has improved dramatically over the last four decades, the prognosis for those who relapse remains dismal, especially for those who relapse while on therapy. In fact, relapsed disease remains a leading cause of cancer related mortality in children. To date, various studies have discovered a number of somatic alterations that contribute to driving relapse and have provided profound insight into the selective forces that lead to clonal outgrowth of drug resistant populations, however these lists are not yet comprehensive.

We analyzed 13 pediatric ALL patients treated according to Nordic NOPHO ALL protocols and explored a comprehensive collection of germline, diagnosis, relapse, and maintenance samples. Whole exome sequencing (WES) was performed on all available germline, diagnosis, and relapse samples to find somatic missense mutations enriched in the relapse samples versus the diagnosis and/or germline samples. Sequencing reads were aligned to the human genome (build hg19/GRCh37) using the Burrows-Wheeler Aligner (BWA) and single-nucleotide somatic variants were generated with MuTect. ANNOVAR was used to annotate variants with functional consequences and identify if the variant was contained in dbSNP, ExAC, 1000 Genomes project, and COSMIC databases. Nine of the NOPHO patients were analyzed as trios (WES of germline, diagnosis, and relapse), three of the patients as Diagnosis-Relapse duos and one as a Germline-Relapse duo. Candidate relapse driving mutations were identified as present at high levels in the relapse sample, but were undetectable in germline or low to absent in the diagnosis sample. Missense mutations had to be enriched by ≥5% in the relapse sample versus diagnosis/germline to be included for further consideration. Relapse specific candidates were further prioritized based on tumor percentage (≥ 20%), bioinformatic tools predicting a missense change being deleterious or damaging to protein function, and literature reviews for insight into the biological pathway potentially affected.Eight of the thirteen patients contained mutations in genes previously reported to be enriched and are involved in nucleoside metabolism/synthesis, histone acetylation, transcription regulation, or cell signaling/growth through the Ras pathway. Interestingly, a majority of the patients contained novel relapse specific genes in a major clone that met the criteria for drivers (Table 1). These novel candidates are involved in a wide array of cellular processes such as cell adhesion/migration, RNA polymerase II/transcription, circadian rhythm, the unfolded protein response, RNA transport, epigenetic regulation, DNA methylation, and kinases.

Knowing the exact relapse specific mutations for each patient allows use of droplet digital PCR (ddPCR) to track the emergence of specific candidate mutations from peripheral blood samples (range of 2-68 per patient, Table 1) collected from these patients prior to relapse. Thus far, we have successfully backtracked the emergence of the NT5C2 p.R367Q mutation (.2% Minor Allele Frequency (MAF)) just over a month before frank relapse in patient 8142, using ddPCR. Tracking these mutations offers insight into which mutations drive relapse and the speed at which the relapse clones emerge. Probes for ddPCR to detect our top candidates have been developed and are currently being applied. Ultimately, candidate mutations emerging with the major clone will undergo functional testing to understand the mechanism by which the mutation drives relapse. Through these approaches, we will be able to pinpoint what mutation(s) and combinations thereof drive relapse through clonal survival during maintenance therapy.


No relevant conflicts of interest to declare.

Author notes


Asterisk with author names denotes non-ASH members.

Sign in via your Institution