Abstract
Most mutations in AML are random, benign events that accumulate with age in HSPCs.
Major finding: Most mutations in AML are random, benign events that accumulate with age in HSPCs.
Approach: Whole-genome sequencing was performed on AMLs with known and unknown initiating events.
Impact: Mutations in NPM1, DNMT3A, and IDH1 are likely initiating events in normal-karyotype AML.
Most somatic mutations identified in cancers are assumed to have occurred after the tumor-initiating mutation, which may promote genomic instability and the acquisition of additional mutations, but the evolutionary process of genomically stable cancers is less clear. To gain insight into the evolution of acute myeloid leukemias (AML), approximately half of which have normal karyotypes, Welch and colleagues compared the genomes of 12 M3-subtype samples initiated by the PML-RARA gene fusion with 12 M1-subtype samples with a normal karyotype in which the initiating event was unknown. Hundreds of single-nucleotide variants (SNV) were identified in nearly all the cells within both subtypes, and the number of total mutations increased with patient age. However, the number of mutations in normal hematopoietic stem/progenitor cells (HSPC) also increased as a function of age and was similar between age-matched healthy volunteers and AML patients, suggesting that most of the mutations identified in the AML samples represented random preexisting events in the HSPC that had clonally expanded after an initiating mutation. Nonsynonymous mutations were found in an average of only 10 to 11 genes per sample, with a median of 2 to 3 of these found in multiple AML genomes. Nine genes, including FLT3, were recurrently mutated in both subtypes, and thus may represent cooperating events that can interact with different initiating events. Thirteen genes were recurrently mutated only in M1 AML, including NPM1, DMNT3A, and IDH1, which each had significantly higher mutation frequencies in M1 samples than in M3 samples and therefore may represent initiating events in the M1 subtype. Together, these findings indicate that the vast majority of mutations predate the cancer-initiating event in AML, and that relatively few mutations are required for the initiation and progression of these cancers.