RNA Sequencing Illustrates the Genetic Backgrounds of Pediatric Acute Myeloid Leukemia

Background

The cytogenetic heterogeneity of acute myeloid leukemia (AML) has been recognized for more than three decades, and some chromosomal aberrations and gene mutations have been adopted for the risk stratification. However, approximately 50% of patients have disease relapse. Thus, novel and accurate prognostic markers are needed to improve the risk stratification. We have performed some targetting sequencing and gene expression array, and identified that PR DM16 and MECOM -overexpressing patients had significantly worse overall survival (OS) and event-free survival (EFS). Thisoverexpression was especially helpful for stratifying patients with FLT3 -ITD positive and/or normal karyotype. To establish a more appropriate risk classification, we investigated the correlations among genetic aberrations, gene fusions, and gene expression, which are significant prognostic markers for pediatric AML, and proposed a new risk stratification. In addition, we performed the whole-transcriptome analysis to reveal the comprehensive understanding of 57 patients whose aberrant genetic backgrounds were unclear.

Patients and Methods

This study enrolled 369 patients with de novo AML who had been registered with the Japan Children's Cancer Group study, JPLSG, AML-05 trial between November 2006 and December 2010. Patients diagnosed with acute promyelocytic leukemia or Down syndrome-associated AML were excluded from this study. We investigated gene mutations in KIT, CEBPA, NPM1, FLT3 -ITD, and MLL- PTD, gene fusions of RUNX1-RUNX1T1, CBFB-MYH11, KMT2A (MLL)- rearrangements, NUP98-NSD1, CBFA2T3-GLIS2, NUP98-KDM5A, FUS-ERG, and DEK-NUP214, and gene expression analyses of PRDM16 and MECOM . Although most of these patients were classified into five risk-dependent subgroups based on their genetic aberrations, the genetic background of 57 patients (15%), particularly in patients with complex karyotypes, was unclear. Thus, we additionally performed the whole-transcriptome analysis for these patients to reveal their genetic backgrounds.

Results

We could re-stratify 369 patients into five new prognostic subgroups based on the genetic aberrations. Because genetic backgrounds of patients with acute megakaryocytic leukemia (AMKL) were different from non-AMKL patients, we proposed another stratification scheme for 34 patients with AMKL. For 335 patients with AML, as the first step, super low and extreme high groups were created. The super low group (5-year OS, 94%; 5-year EFS, 84%) consisted of CBFB-MYH11, KMT2A-R with low MECOM expression, NPM1 mutation, and CEBPA biallelic mutations. Conversely, the extremely high group (5-year OS, 24%; 5-year EFS, 16%) consisted of FUS-ERG, MLL-AF6, DEK-NUP214, CBFA2T3-GLIS2, MLL -PTD, NUP98-NSD1, MLL -R with MECOM and/or PRDM16 high, and FLT3 -ITD with high PRDM16 gene expression. Then, we chose low and intermediate groups as the second step. The low group (5-year OS, 86%; 5-year EFS, 76%) consisted of RUNX1-RUNX1T1 without KIT, MLL -R with low MECOM and PRDM16, and the intermediate group (5-year OS, 85%; 5-year EFS, 58%) consisted of RUNX1-RUNX1T1 with KIT, FLT3 -ITD with low MECOM, and PRDM16, according to the retrospective EFS and OS. For the remaining 57 patients (15%) whose genetic background was unclear, non-complete remission (CR) patients were assigned to the extremely high group, and patients who achieved CR were assigned to the high risk group (5-year OS, 53%; 5-year EFS, 44%). We performed the whole-transcriptome analysis for these 57 patients and identified gene fusions of PICALM-MLLT10 in 3, RPN1-MECOM in 2, CREBBP-KAT2A in 3, RUNX1 -rearrangements in 2, CBFA2T3 -rearrangement in 2, RARB -rearrangement in 1, and ETV6 -rearrangements in 3. The whole-transcriptome analysis was used to reveal the genetic background of these patients, particularly in patients with a complex karyotype. For the remaining 34 patients with AMKL, patients with CBFA2T3-GLIS2 or NUP98-KDM5A were assigned to the extremely high group and others to the intermediate group.

Conclusion

We propose a new risk stratification comprising five prognostic subgroups based on the genetic aberrations. The whole-transcriptome analysis facilitated further understanding of the whole genetic background of pediatric patients with AML whose genetic background was unclear and was, thus, expected to improve more accurate risk stratification.