Flow cytometry reveals the nuances of clonal haematopoiesis

Abstract

Advances in genetic sequencing technologies in the past decade have led to our deeper understanding of the molecular ontogeny of acute myeloid leukaemia (AML). Mutated haematopoietic clones [clonal haematopoiesis (CH)] appear to precede most AML cases and may or may not be associated with cytopenias or overt myelodysplastic syndrome (MDS). While this concept sheds light on our understanding of the molecular aetiology of myeloid neoplasia, it also creates an important practical issue during genetic interrogation of AML patients following induction chemotherapy: CH often persists, even in AML patients who appear to be morphologically free of disease. With increasing therapeutic options in AML, detecting low levels of persistent leukaemia after therapy [minimal residual disease (MRD)] is becoming a critical part of AML patient management. Multiple studies have shown that patients with MRD have higher risk of relapse compared to patients with no detectable MRD. Moreover, earlier or more intensive therapy may improve the outcome of patients who harbour MRD.1 However, true MRD must be distinguished from CH, as the latter is pre-leukaemic condition that does not necessarily confer an increased risk of relapse.2 While accurate measurement of MRD is rapidly attaining standard-of-care status in AML and is currently recommended by the European Leukemia Net (ELN), its application is not always straightforward. Only certain AML subtypes, such as those with NPM1 mutation or translocations involving core-binding factor genes, are amenable to highly sensitive polymerase chain reaction (PCR)-based genetic MRD detection. Next-generation sequencing (NGS) can be used to identify AML-associated mutations such as NPM1 after treatment, but most current NGS methods are not sufficiently sensitive to the recommended MRD detection level of 1 x 10-3. NGS often identifies mutations in other genes such as DNMT3A, TET2, and ASXL1, which indicate a state of CH that is not considered to be equivalent to true AML MRD.2, 3 For these reasons, the ELN also recommends flow cytometry immunophenotyping to monitor AML MRD post-therapy, since it is applicable to most AML cases. Blasts displaying an immunophenotype associated with the specific patient’s AML, or blast populations that deviate from normal expected immunophenotypic patterns, can be detected with high sensitivity and are considered to represent MRD. However, flow cytometric MRD assessment is technically complex and subject to pitfalls in interpretation, limiting its widespread application at the current time. In this issue of British Journal of Haematology, Loghavi et al.4 describe abnormal myeloid progenitor immunophenotypes (termed ‘pre-leukaemic’ immunophenotypes) detected by flow cytometry in patients with NPM1-mutated AML in remission following treatment. Unlike typical NPM1-mutated AML blasts, these pre-leukaemic myeloid progenitors were CD34+ blasts that showed significantly increased CD123 and/or CD117 expression, frequently with decreased CD38 and/or HLA-DR expression compared to normal myeloid progenitors, in fact resembling blasts found in cases of low-grade MDS. These pre-leukaemic blasts were identified in nearly half of the patients lacking any genetic evidence of residual NPM1-mutated leukaemia. The pre-leukaemic blasts were strongly associated with the presence of background CH (particularly mutations in IDH2 and SRSF2), as well as with high variant-allele frequency (VAF) CH mutations and morphologic dysplasia in the maturing haematopoietic elements. While this might have suggested the presence of background MDS, there was no association of pre-leukaemic immunophenotypes with decreased blood counts compared to patients lacking this finding; moreover, NPM1-mutated AML is almost never associated with an antecedent MDS. This finding was unlikely to be explained merely by marrow regeneration following chemotherapy, since patients receiving the same chemotherapy regimens but lacking CH did not show the same pre-leukaemic-type aberrations. These results are important for two main reasons: (i) although the aberrant blast phenotypes differed from those associated with true residual leukaemia, applying a ‘deviation from normal’ approach in flow cytometry MRD assessment could lead to their misidentification as MRD. Thus, careful attention to the specific blast phenotypes is important in establishing MRD by flow cytometry in the setting of NPM1-mutated AML. Further study is warranted to determine if similar pre-leukaemic blast populations can also be seen in other AML subtypes following treatment. (ii) The aberrant blast phenotypes were associated with specific patterns of CH, in particular high VAF and frequent IDH2 and SRSF2 mutations. Although the authors did not find any differences in relapse during their relatively short follow-up period, their results suggest that some types of CH in the post-AML setting may have MDS-like immunophenotypic features. At the current time, the significance of CH in the setting of AML in morphologic and molecular remission is controversial; the results of the Loghavi study suggest that CH in this setting is variable and that flow cytometry represents an additional tool to study its nuances and clinical implications. Recently, aberrant blast immunophenotype (CD7 expression) and CH were both found to confer increased risk of therapy-related myeloid neoplasia in patients undergoing autologous cell transplantation, underscoring the complementary nature of genetics and immunophenotype in evaluating for abnormal haematopoiesis.5 Further study with longer follow-up and on broader AML cohorts is needed to determine if immunophenotypically abnormal CH in remission may identify patients at higher risk of long-term relapse or other complications that can be associated with CH, such as cardiovascular disease.