Clonal Evolution in Relapsed Pediatric Acute Lymphoblastic Leukemia

Introduction
Clinical outcome of relapsed pediatric B-cell progenitor acute lymphoblastic leukemia (BCP-ALL) remains poor, although survival rate for children with BCP-ALL has greatly increased over time and is now reached 90%. To clarify the molecular pathogenesis of relapsed ALL may provide novel prognostic markers and therapeutic targets. Some genome-wide analyses for specific patients group with poor prognosis, such as early-relapsed patients and Ph- or BCR-ABL-like patients, were reported. They described important insights to understand genetic background of poor prognosis. However, the majority of relapsed cases did not have any poor prognostic marker, and the molecular mechanisms of relapse in these cases still remained unclear. Therefore we performed whole exome sequencing (WES) to describe clonal evolution in 21 relapsed pediatric BCP-ALL patients. Our cohort included various cases whose time to relapse from diagnosis were between 6 months to over 10 years. We also analyzed the clonality of leukemia cells using immunoglobulin (Ig) and T-cell receptor (TCR) rearrangements.
Patients and Methods
Genomic DNA was isolated from 21 cases whose median time to relapse was 33 months. Somatic mutations including SNVs (single nucleotide variants), insertions / deletions and CNVs (copy number variants) were detected by WES using Agilent SureSelect and illumine HiSeq systems. To evaluate accurate VAF (valiant allele frequency), targeted deep sequencing was performed in candidate somatic mutations. The clonality analysis of leukemia cells was performed by standard PCR methods using Ig and TCR rearrangements.
Results
WES was performed in samples obtained at diagnosis, remission and relapse from 21 pediatric BCP-ALL patients. Tumor specific mutations had been identified by WES. Three of 21 were hypermutated with over 150 somatic mutations at relapse. Mutation of DNA mismatch repair gene, MSH3, was detected in 2 of them. Except for these hypermutated cases, the median number of somatic mutations detected at relapsed phase was 22 （range 8 to 53）, which was higher than that at diagnosis (median 16, range 6 to 31). Sixteen recurrently mutated genes were identified in 21 cases by WES. Some known leukemia associated genes were detected, including KRAS and WHSC1 observed only at diagnosis and IKZF1 and CREBBP observed at relapse. Then we compared VAFs of these mutations between at diagnosis and relapse to solve the clonal architectures over time. Three patterns of clonal evolution were estimated from VAFs using targeted deep sequencing; (i) In 7 cases, all mutations described at diagnosis were shared at relapse, suggesting that relapse clone derived from predominant clone at diagnosis with additional mutations in these cases. (ii) In other 13 cases, most of mutations in predominant clone at diagnosis were not detected at relapse except for some shared mutations at diagnosis and relapse, indicating that relapsed clone occurred from founder clone existing as subclone at diagnosis. (iii) In one very late-relapsed case, there were no shared mutations at diagnosis and relapse. According to clonality analysis of Ig and TCR, none of rearrangements identified at diagnosis were conserved at relapse in this case. On the other hand, most rearrangement at diagnosis were conserved at relapse in other 20 cases except one patient who relapsed in 10 years after diagnosis. Relapse from predominant clone at diagnosis were observed in only one out of 8 late-relapsed cases (> 36 months), whereas a half of the early-relapsed showed this clonal evolution pattern. The number of shared mutations between diagnosis and relapse was very limited in very late-relapsed cases over 10 years.
Discussion
Our study suggests that the clonal evolution pattern differs according to the time to relapse. In a half of early-relapsed cases, relapsed clone derived from major clone at diagnosis with additional mutations, and clonal selection of resistant clones occurred during treatment. Meanwhile, in late-relapsed cases, relapse was frequently associated with clonal evolution from minor subclone with some conserved mutations and same Ig/TCR rearrangements. The founder clone should be remained dormant for a long period until additional mutations lead to relapse. Towards a better understanding of clonal evolution in ALL, our study will shed light on the early prediction of relapse risk and new treatment strategies for relapsed ALL.