Clinical Effect of Genetic Alterations in Pediatric Patients with B-Progenitor Acute Lymphoblastic Leukemia

Background

Recent genetic studies using high-throughput sequencing have substantially revealed genetic alterations in pediatric patients with B-progenitor acute lymphoblastic leukemia (B-ALL). However, their clinical relevance of these alterations has not been fully investigated.

Methods

We performed targeted-capture sequencing of 149 known or putative driver genes for B-ALL in 525 pediatric patients who were uniformly treated according to the Japan Association of Childhood Leukemia Study ALL-02 protocol. In addition, we performed whole transcriptome sequencing in 63 of 118 patients who were not categorized into any genetic subgroup defined by RT-PCR status or targeted-capture sequencing results.

Results

In total, 647 focal deletions (median 1 per patient; range, 0-7) and 993 mutations (median 2 per patient; range, 0-10) were detected in 491 (94%) cases. Some of these alterations were closely associated with disease subtypes. We identified recurrent PHF6 mutations, which were almost exclusively detected in the TCF3 - PBX1- positive ALL genotype. Furthermore, frequent involvement of PAX5 and KMT2D (mutations) and IKZF3 (deletions) were characteristic of TCF3 - PBX1 -positive ALL genotype. To identify genetic alterations associated with a poor prognosis, we analyzed 20 genes that were mutated in at least 20 patients in our study cohort. Univariate analyses revealed genetic alterations in TP53 , KMT2D , and IKZF1 to be associated with a significantly shorter overall survival. Mutations and deletions in IKZF1 were also strongly associated with induction failure (Odds ratio, 12; 95% CI, 2.7-55). This led to death in 80% (n = 4/5) of the patients with IKZF1 alterations. Most patients with TP53 and KMT2D mutations and with disease relapse had fatal disease progression [94% mortality rate (n = 17/18)]. Multivariate analysis, together with the NCI criteria, revealed that two genetic lesions, TP53 and KMT2D , retained prognostic impacts for overall survival. Prediction of 5-year OS improved in each NCI-risk group when patients in each risk group were stratified according to the presence or absence of mutations in TP53 or KMT2D lesions: 69% [with alteration (n = 26)] vs. 85% [without alteration (n = 138)] for the NCI-HR group (P = 0.03) and 85% [with alteration (n = 42)] vs. 97% [without alteration (n = 319)] for the NCI-SR group (P = 0.004). Additionally, the four less frequently mutated genes also had an independent prognostic impact before and after adjustment for risk factors: IKZF2 (n = 6; HR for death, 7.4; 95% CI, 2.3-24), SH2B3 (n = 4; HR for death, 7.0; 95% CI, 1.7-29), JAK2 (n = 7; HR for events, 4.6; 95% CI, 1.7-12), and PTEN (n = 9; HR for death, 3.8; 95% CI, 1.2-12). Notably, mutations in three genes present potentially druggable targets: ruxolitinib for mutations in JAK2 or SH2B3 and poly (ADP-ribose) polymerase inhibitor formutations in PTEN . Finally, based on a gene expression profile, 20 of 63 patients were categorized into discrete subgroups characterized by Ph-like profiles (n = 7), ZNF384 (n = 7), and MEF2D (n = 6) fusions. Ph-like and MEF2D fusion-positive ALLs were associated with a dismal prognosis, whereas all patients with a ZNF384 fusion survived with complete remission. No mutations were associated with a poor prognosis in the 43 remaining patients. Further analysis, such as whole genome sequencing, is warranted for these patients.

Conclusions

Targeted-capture sequencing improves the prediction of clinical outcomes and helps detect prognostic and druggable biomarkers. Our results can facilitate the clinical applications of next-generation sequencing to select optimal therapy for pediatric patients with B-ALL.