Genomic Integration of Adult and Pediatric Acute Myeloid Leukemia Reveals Age Dependent Risk Association

Adult and pediatric acute myeloid leukemia (AML) are clinical entities separated by an arbitrary age cutoff, yet they share substantial cytogenetic and morphological features. A 21-year-old newly diagnosed AML patient with a KMT2A-MLLT3 translocation who presents at an adult hospital will be offered a stem cell transplant in first remission if a suitable donor is available based on European LeukemiaNet (ELN) risk stratification whereas a pediatric facility will treat with chemotherapy alone provided they achieve MRD negative status after two cycles.

To understand whether age defines molecular groups or alternatively informs risk, we compiled a cohort of 1307 clinically and molecularly well characterized AML patients with long-term follow-up who were similarly treated with anthracycline-based induction chemotherapy (pediatric patients, n=435, adult patients, n=872, median age 38 [range: 0-84 years]). Biological diversity of AML is best captured by gene expression signatures of highly variant genes. We therefore used a previously characterized gene expression classifier to integrate the pediatric and adult cohorts, yielding subsets with mixed adult and pediatric composition. Notably, a previously unrecognized pediatric subset merged with one of the dominant adult subsets, characterized by mutations present in myelodysplastic syndrome (MDS) and secondary AML (sAML) such as RUNX1, ASXL1/2, TP53, DNMT3A, TET2 and SRSF2. Males were represented at a higher frequency in this subset (p=.006), consistent with the higher proportion seen in the clonal hematopoiesis, MDS, and sAML clinical entities. Additionally, clinically confirmed adult sAML patients from a previously published cohort clustered with this subset, indicating that AML in these patients may have developed from a subclinical myelodysplastic state. Other major iso-transcriptional subsets with a pediatric to adult composition include RUNX1-RUNX1T1, KMT2A rearrangements, CBFB-MYH11, CEBPA, Acute Myeloid and T-Cell precursor Leukemia (AMTL) and a subset with high HOX cluster gene expression characterized by NPM1 mutations, nucleoporin rearrangements and/or FLT3 internal tandem duplications (ITD). To understand the association of age and risk, we analyzed an expanded cohort of 2217 AML patients that belonged to one of 8 molecular subsets: CEBPA, CBFB-MYH11, RUNX1-RUNX1T1, KMT2A-MLLT3, NPM1 single mutant, FLT3-ITD single mutant, NPM1/FLT3-ITD double mutant, and MDS-like (pediatric patients, n=790, adult patients, n=1427, median age 36 [range: 0-59 years]). Treating age as a continuum, we found sharp or gradual decreases in overall survival, depending on AML subset. An optimal cut algorithm was used to identify age cut-offs that distinguish good and poor risk prognoses within molecular subgroups which varied significantly. Patients with KMT2A-MLLT3 that were greater than 12 years of age were identified as having an inferior prognosis whereas age greater than 45 years was associated with poor outcomes for FLT3-ITD positive patients (Figure A, B).

Mutational analysis of the entire cohort revealed a higher burden of cooperating mutations in certain molecular subsets in the older risk groups. For example, older patients carrying NPM1 mutations in the absence of FLT3-ITD had a higher occurrence of DNMT3A and single CEBPA mutations. In CEBPA double mutated, KMT2A-MLLT3 and CBFB-MYH11 subsets, significant age risk associating variables included an increased transcriptional resemblance to fetal liver hematopoietic progenitor cells, suggesting an additional age dependent role of quiescence potential in disease outcome. Age risk thus may be a function of both mutational timing and oncogenic driver lesion.

In conclusion, we propose that pediatric and adult AML can be integrated, with common biological subsets in which age plays different roles. Both pediatric and adult clinical practice of treating AML may benefit from this approach of age integrated analysis.

M.F. and D.N. contributed equally. Co-corresponding authors: A-K.E. T.A.G.

Support: U10CA180821, U10CA180882, U24CA196171, https://acknowledgments.alliancefound.org. Clinicaltrials.gov Identifiers: NCT00048958 (CALGB 8461), NCT00899223 (CALGB 9665), NCT00900224 (CALGB 20202).

Moore:Pharmacyclics: Speakers Bureau. Stone:Epizyme: Consultancy; Astellas: Consultancy; GSK: Consultancy; Innate: Consultancy; OncoNova: Consultancy; Foghorn Therapeutics: Consultancy; Jazz: Consultancy; Elevate Bio: Consultancy; Boston Pharmaceuticals: Consultancy; Gemoab: Consultancy; Apteva: Consultancy; Janssen: Consultancy; Syndax: Consultancy; BMS: Consultancy; Novartis: Consultancy; BerGenBio: Consultancy; Arog: Consultancy, Research Funding; Kura Oncology: Consultancy; Syntrix: Consultancy; Aprea: Consultancy; Syros: Consultancy; Takeda: Consultancy; Abbvie: Consultancy, Research Funding; Actinium: Consultancy. Stock:Servier: Honoraria; Pluristem: Consultancy, Honoraria; Pfizer: Consultancy, Honoraria, Research Funding; MorphoSys: Honoraria; Kura Oncology: Honoraria; Kite: Honoraria; Jazz Pharmaceuticals: Honoraria; Amgen: Honoraria; Agios: Honoraria; Syndax: Consultancy, Honoraria; Newave Pharmaceuticals: Consultancy. Byrd:Kura Oncology, Inc: Consultancy; Janssen Pharmaceuticals, Inc.: Consultancy; Syndax: Consultancy; TG Therapeutics: Honoraria; Vincerx Pharma: Current equity holder in publicly-traded company; Pharmacyclics LLC: Honoraria, Research Funding; AstraZeneca: Consultancy; Xencor, Inc: Research Funding; Novartis: Consultancy, Honoraria. Eisfeld:Karyopharm Therapeutics: Other: Spouse is current company employee. Gruber:Kura Oncology: Consultancy.