Biology and Prognostic Markers in Young Children with Acute Myeloid Leukemia -the Jccg Study, JPLSG AML-05-

[Introduction] Acute myeloid leukemia (AML) in children aged < 12 months of age was conventionally defined as infant AML. On cytogenetic and molecular analysis, 11q23/ KMT2A rearrangement (KMT2A -R) was found to be the most frequent (approximately 50%) cause of infant AML. On prognostic analysis, infant AML was reported to have favorable outcome compared with AML in older children, and only KMT2A -R and RBM15-MKL1 were known to be intermediate-risk prognostic factors in infants. Although infants have been treated in the same regimens as older children, in some clinical trials, drug dosage for infants has been reduced to avoid severe toxicities. However, new findings in young children with AML have been recently accumulated. A recent report by an international expert panel of pediatric AML has proposed AML in children aged < 24 months of age as a distinct age-subgroup because their morphologic and cytogenetic/molecular characteristics were considered to be similar to those of infants (Creutzig et al, 2012).　Moreover, recent studies have identified novel inv(16)(p13.3q24.3)/ CBFA2T3-GLIS2 and t(11;12)(p15;p13)/ NUP98-KDM5A fusion genes in AMKL. In this study, we investigated biology and prognosis of children aged < 5 years, to identify unique characteristics of young children with AML and propose a novel risk-stratification for them.

[Patients and Methods] A total of 369 children with AML were enrolled in the Japanese Pediatric Leukemia/Lymphoma Study Group (JPLSG) AML-05 trial. In this study, we analyzed 133 children aged < 5 years (36%). We screened fusion genes of CBFA2T3-GLIS2, NUP98-KDM5A, RBM15-MKL1, KMT2A -Rs, NUP98-NSD1, FUS-ERG, DEK-NUP214, RUNX1-RUNX1T1, and CBFB-MYH11, and gene mutations of KIT, NRAS, KRAS, WT1, NPM1, KMT2A -PTD, FLT3 -ITD, CEBPA, ASXL1, ASXL2, and CSF3R .

[Results] In clinical characteristics, no change was observed in the sex ratio with age. The WBC count at diagnosis was the highest in children aged 2 years (24-35months), and a trend of lower WBC in children aged > 2 years (> 35 months). On FAB classification, the proportion of M5 and M7 was observed to be decline with age. Among fusion genes, KMT2A -R was the most frequently identified fusion gene (23%), followed by RUNX1-RUNX1T1, CBFB-MYH11, and CBFA2T3-GLIS2 (13.5%, 9.8%, and 8.3%, respectively). Conversely, neither NUP98-NSD1 nor DEK-NUP214 was found. Although the frequencies of KMT2A -R and CBFA2T3-GLIS2 tended to decrease with age, the frequency of RUNX1-RUNX1T1 significantly increased with age. Among gene mutations, KIT and NRAS were the most frequent, and gene mutations other than ASXL1 and KMT2A -PTD were recurrently identified. When correlation of fusion genes with gene mutation was analyzed, 21 (68%) of 31 children with CBF-AML had gene mutations, whereas only 8 (16%) of 51 children with non-CBF-AML had gene mutations (p < 0.001). For fusion genes, 28 (90%) of 31 KMT2A -R-positive children were aged < 3 years (< 36 months) (90%), and 17 (94%) of 18 RUNX1-RUNX1T1 -positive children were aged > 2 years. Thus, according to these molecular characteristics of young children with AML, we defined children aged < 3 years as the youngest subgroup with distinct biology (n = 99). We further performed prognostic analyses of the youngest subgroup. According to recent previous studies that reported poor prognosis of CBFA2T3-GLIS2 and NUP98-KDM5A in AMKL, these fusions were combined to the same group. The Kaplan-Meier analyses revealed that CBFA2T3-GLIS2 / NUP98-KDM5A -positive children had the poorest outcome in OS, EFS, and CIR (38%, 13%, and 88%, respectively). Surprisingly, children with CBF-AML had the second poorest outcome in EFS and CIR (45% and 55%, respectively), although this subgroup had an excellent OS (100%), indicating that children with CBF-AML in the youngest subgroup have a high-risk of relapse and might need more intensified therapy as an initial treatment.

[Conclusions] The youngest subgroup of pediatric AML had a unique biology. Children with CBFA2T3-GLIS2 and NUP98-KDM5A had the poorest outcome, and children with CBF-AML also had a high relapse rate. Because this subgroup is known to have a high incidence of severe side effects and late effects, a novel risk-stratified therapy restricted to young children is required.