Genomic Classification and Prognosis in Acute Myeloid Leukemia

To be a doctor is to classify. Perhaps the two major challenges of medical college are firstly, to understand how to talk to patients and their relatives, and secondly, to learn a long series of diagnostic lists. However, as with everything in life, a balance must be struck. An obsession with excessive categorization has itself been regarded as a medical problem. So where does this leave us in hematology?

When John Hughes Bennett noticed the disorder of “white blood” in 1845, he imagined leukemia to be a single condition. The work of Dr. William Dameshek and colleagues established a classification from which the subdivision of acute myeloid leukemia (AML) emerged. For many years, AML was subdivided on the basis of morphology, aided over time by cytochemistry and flow cytometry. However, the realization that cancer develops from the accumulation of somatic mutations predestined a dominant role for genomic architecture within disease classification.

The current World Health Organization (WHO) classification defines AML largely on the basis of mutational analysis. In this report, Dr. Elli Papaemmanuil and colleagues extend this profile by combining clinical and cytogenetic information with targeted sequencing of 111 candidate driver genes in 1,540 patients. Their conclusion is that 11 largely discrete subtypes of AML can be distinguished. The patients were mostly younger patients undergoing intensive therapy, enrolled in trials of the German-Austrian AML study groups. The first interesting observation was on the nature of the genetic damage in AML. Driver mutations included point mutations, gene deletions or insertions, recurrent fusion genes, and aneuploidies. Seventy-six such drivers were defined within the cohort, and at least one driver was observed in 96 percent of cases, with 86 percent of tumors carrying two or more genetic hits. Further insight was gained into the natural history of hemato-oncology disorders. For example, mutations in genes that encode epigenetic modifiers, such as DNMT3A, ASXL1, IDH1/2, and TET2, occurred very early during clonal evolution but were almost never enough on their own to cause leukemia. Secondary mutations such as NPM1, JAK2, or SF3B1 seem to be required to drive these clones to AML, myeloproliferative disease, or MDS, respectively.

Reassuringly, the current WHO subgroups held up in this study, but a major finding was the identification of three novel genetic subgroups. These comprised mutations in genes encoding chromatin, RNA-splicing regulators, or both in 18 percent of patients; TP53 mutations, chromosomal aneuploidies, or both in 13 percent; and an uncommon subgroup of IDH^R172 mutation in 1 percent. Importantly, only 48 percent of patients were classifiable based on current WHO guidelines whereas this novel approach increased this allocation to 80 percent.

Perhaps disappointingly, this huge increase in genetic knowledge did not greatly improve prediction of overall survival, which was possible in 71 percent of cases. This reflects some limitation of focusing solely on the mutational complexity of the tumor, as the integrity and response of the host, including immune function, will be an important determinant of outcome.

The information will probably not lead to a major change in treatment, though identification of rare subgroups with mutations in targetable genes such as BRAF may offer an insight into greater stratification.