The complex genetic landscape of acute myeloid leukaemia

Abstract

Classification of acute myeloid leukaemia (AML) has changed substantially since the French–American–British (FAB) system was established in the 1970s. The system—which divided the disease into seven morphological subtypes—was largely based on histology, and reflected the extent of differentiation of the blast cells. Although most subtypes were treated in the same manner, the FAB classification saw a breakthrough in terms of disease-specific treatment for patients with the M3 subtype, acute promyelocytic leukaemia. Although the M3 subtype was classified by histology, the discovery of the specific underlying chromosomal translocation containing the PML-RARA fusion gene showed the link between morphology and genetics in this cancer. Use of all-trans retinoic acid to promote differentiation transformed this subtype into one that is frequently curable. As the genetic landscape of AML has been further uncovered, classification systems have become more refined but increasingly complex. In the 2000s, WHO defined eight broad categories of AML, with further subgroups within those categories. Chromosomal aberrations and translocations make up most of the classifications based on genetics, and notably, the category of “acute myeloid leukaemia, not otherwise specified” is a direct transposition of the FAB classification system. Alongside classification of leukaemias from a diagnostic point of view, greater insight into how genetics affect outcome has led to the identification of four distinct risk groups by the European LeukemiaNet Standardized Reporting System (ELN), which have subsequently been shown to improve prognosis and guide treatment decisions. Against this background of ever further refinement, data for the genetic landscape of 1540 patients with AML were reported by Elli Papaemmauil and colleagues on June 9. They were able to identify patterns of co-mutation from 5234 driver mutations across 76 genes and genomic regions that allowed patients to be subdivided into 11 distinct classes. Over half of the patients genotyped could be grouped into the three most common genomic subgroups. Only 15% of patients had either no driver mutation detected or that could not be classified into one of the 11 subtypes. Although mutational studies involving large cohorts of patients with AML and with other malignant conditions are useful for understanding leukaemogenesis and potentially improving patient risk stratification, whether these subtypes are robust enough in other patient cohorts to further subdivide patients into distinctive risk and treatment categories is unknown. Currently, most patients with AML are treated with regimens established in the 1970s and treatment recommendations are largely based on baseline characteristics, such as age. Understanding the genetics of this heterogeneous disease will hopefully lead to a new era of targeted and disease-specific therapies. Since the discovery of FLT3 mutations in AML, there has been a flurry of research activity to try to develop an effective inhibitor, with several clinical trials in progress. Other mutations, such as those in NPM1, do not lend themselves to simple drug targeting. However, such mutations could provide a sensitive target for minimal residual disease monitoring. When the pathways are further elucidated, individuals could be screened for the risk of developing leukaemia. Already, the genetic changes in this recent AML genetic study are blurring the lines between high-risk myelodysplastic syndromes and AML, with the same changes being seen in both diseases. WHO are updating the classification systems for lymphoma and leukaemia. With the genetic information now available, is it perhaps time to remove reference to the morphology of the cells altogether? When the WHO classification was first introduced, about half of AMLs had a normal karyotype and could potentially be classified in the not otherwise specified category; however, only 4% of the patients in Papaemmauil and colleagues' study had no mutation identified, indicating that morphology is perhaps redundant. That said, the morphology of the blast cells reflects their underlying genetics and could potentially still provide insight into disease treatment. As the classification of the disease becomes increasingly complex, what is needed now is the identification of key tests that can help best stratify patients into risk groups and guide treatment. Not all hospitals and cancer centres, particularly in low-income and middle-income countries, will have the facilities to perform the full gauntlet of genetic assays and flow cytometry to fully classify the disease, although this should be a goal for all facilities to aim towards. A consensus now needs to be reached to identify which key diagnostic tests are needed in an ever more complex landscape.