Landscape of Driver Mutations and Their Clinical Impacts in Pediatric Acute Lymphoblastic Leukemia

Introduction

B-progenitor acute lymphoblastic leukemia (B-ALLs) accounts for 85% of pediatric ALL and categorized into several molecular subgroups according to their ploidy and recurrent translocations, such as ETV6-RUNX1, TCF3-PBX1, BCR-ABL1, and MLL-rearrangements. In addition, recent genetic studies using high-throughput sequencing have disclosed landscapes of gene alterations in each subgroup, however, their clinical relevance have not fully been investigated in a large cohort of B-ALL patients who are uniformly treated and enrolled in an unbiased manner.

Methods

We enrolled a total of 515 pediatric B-ALL patients, who had been uniformly treated according to the Japan Association of Childhood Leukemia Study (JACLS) ALL-02 protocol between 2002 and 2008. These patients were categorized into three risk groups, including standard-, high-, and extremely high-risk. Infantile ALL as well as BCR-ABL1-positive and Down syndrome-associated cases were excluded. A total of 158 known or putative driver genes in pediatric ALL were analyzed for somatic mutations by targeted-capture sequencing. IgH rearrangements were captured using 662 baits tiling the entire IgH enhancer locus. Finally, an additional 1205 baits was also designed to enable sequencing-based genome-wide copy number detection.

Results

The median age at diagnosis and observation period were 5.2 (1-18.5) and 4.2 (1.8-9) years, respectively. Sixty-six of the 515 patients (13%) had relapsed diseases and 47 patients (9%) had been died. Real-time RT-PCR and conventional cytogenetic analyses revealed subgroup-defining genetic lesions in 368/515 (71%) patients: 117 (23%) cases with ETV6-RUNX1, 48 (9%) with TCF3-PBX1, 13 (3%) with MLL rearrangements, together with those with hyper- (169 [33%]), and hypo- (6 [1%]) diploid. Remaining 162 patients (31%) had none of these abnormalities. The mean depth of the targeted sequencing was 569× across the entire cohort. In total, 823 driver mutations (median 1 per patient, range 0-7) and 954 focal deletions (median 2 per patient, range 0-13) were detected in 483 patients (92%). Among these, most frequently detected were mutations/deletions in CDKN2A (24%), ETV6 (21%), NRAS (18%), KRAS (18%), and PAX5 (15%). IgH-rearrangements were detected in 51 patients, including IGH-DUX4 (26 [5.0%]), IGH-EPOR (3 [0.6%]) and IGH-CRLF2(2 [0.3%]).

Genetic alterations were enriched in several functional pathways, of which most frequent was epigenetic regulation (53%), followed by B-cell development (47%), RAS signaling (46%) and cell cycle (40%). A number of novel recurrent genetic lesions were also identified, including those in DOT1L and PHF6. DOT1L encode an H3K79 methyltransferase and was inactivated by frameshift/nonsense mutations and/or deletions in 19 cases. Although frequently found in T-ALL, mutations of PHF6 had not previously been reported in B-ALL but were detected in 14 cases in the current cohort and strongly associated with TCF3-PBX1 translocation.

Significant positive correlations were also demonstrated for an additional 10 combinations of common genetic lesions, suggesting functional links between these combinations. Thus, ERG deletions were highly associated with IGH-DUX4 rearrangement, while mutations in KRAS, NRAS, and CREBBP were significantly enriched in hyperdiploid cases. ETV6-RUNX1 fusion also showed positive correlations with alterations in ETV6, CDKN1B, ATF7IP, VPREB1, BTG1, and WHSC1. Furthermore, mutually exclusive relationship between ETV6-RUNX1 translocationsand FLT3mutations were also identified.

Finally, we analyzed the prognostic impact of driver mutations. In multivariate analysis of the entire cohort, 4 genetic alterations were significantly associated with poor prognosis (HR [95%CI]): IKZF1 mutations/deletions (2.6 [1.5−4.8]), EBF1 deletions (3.0 [1.4−6.5]), KDM6A mutations/deletions (2.8 [1.2−6.5]), and TP53 mutations (2.7 [1.2−5.9]). Additional factors (q < 0.1) were identified in subgroup analyses, including alterations in ETV6 (5.4 [1.2−24]), CDKN1B (7.4 [1.6−33]) and CDKN2A (4.2 [1.4−12]) in ETV6-RUNX1 ALL, KMT2D (5.9 [1.3−26]) in TCF3-PBX1 ALLand TP53 (38 [4.1−364]) in IGH-DUX4ALL.

Conclusions

We revealed the landscape of genetic lesions in pediatric B-ALL including novel targets of recurrent mutations with clinical relevance of common genetic lesions. Our results should help in the better stratification of patients.