Immunophenotype analysis is highly helpful for the diagnosis and monitoring of minimal residual disease in hematologic malignancies [3–5]. Though leukemic blasts from NPM1-mutated AML usually show a specific immunophenotype with expression of CD13 and CD33 but absence of CD34 and HLA-DR , different expression patterns of surface markers on leukemic cells from individual patients are frequently seen . In this cohort study, hierarchical cluster analysis revealed two distinct immunophenotypic clusters in NPM1-mutated patients. Most patients in group I showed CD7(−) CD33(+) CD34(−), while almost all patients in group II expressed HLA-DR, CD7, CD33, and CD34 on leukemic cells. The patients in immunophenotypic cluster group II had poorer outcomes than cluster group I, and the immunophenotypic cluster was an independent prognostic factor.
Some AML subtypes showed specific immunophenotypic patterns of leukemic cells, such as coexpression of CD15, CD34 and sometimes, CD19 in AML with t(8;21); coexpression of CD13, CD33 in absence of CD34 and HLA-DR in AML with t(15;17) [4–6]; and coexpression of CD34, HLA-DR, CD15 and CD7 in AML with CEBPA mutation . The prognostic implication of immunophenotype in AML remains controversial [15, 16]. For example, the negative prognostic effect of CD34 expression has been reported in some studies [25, 26], but not in others [27, 28]. The same is also true for CD7 expression. Most studies analyzed the correlation of the expression of a single marker with clinico-laboratory characteristics in a rather heterogeneous group of AML patients. In this study, the immunophenotypic cluster profiles were analyzed in a relatively homogeneous population of AML patients. Immunophenotypic cluster profiles provided distinct prognostic information in NPM1-mutated AML patients.
There are several large studies of NPM1 mutation in AML; the presence of FLT3-ITD is shown to be a poor prognostic factor in NPM1-mutated patients [29, 30]. In order to clarify the association of the immunophenotypic patterns of NPM1-mutated AML with other gene mutations in AML, we checked class I (FLT3-ITD, FLT3-TKD, PTPN11, JAK2, KIT, NRAS, KRAS, and WT1) and class II gene mutations (CEBPA and MLL-PTD). Although the patients in immunophenotypic cluster group II had a higher incidence of FLT3-ITD, a mutation associated with poor prognosis , Cox regression multivariate analysis revealed that the immunophenotypic cluster was an independent prognostic factor (RFS, p < 0.001; OS, p = 0.001) in AML patients with NPM1 mutations. To further evaluate whether the data of flow cytometry could be directly applied for prognostic prediction in clinical practice, we compared the survival between the patients with expression of all HLA-DR, CD34 and CD7 on leukemic cells and other patients. We found that positivity of all three markers was associated with shorter RFS and OS (p = 0.006 and 0.02, respectively). So, it may be worthwhile to use the expression pattern of these three antigens obtained from flow cytometry to predict the survival of patients at diagnosis.
Immunophenotypic cluster in NPM1-mutated AML patients has not been described before. In order to confirm the prognostic effect of the immunophenotypic cluster, we validated the correlation of immunophenotypic cluster and clinical outcome in another cohort of 36 NPM1-mutated patients diagnosed between 2008 and 2010. Hierarchical cluster analysis also showed two distinct clusters, group I patient showed significant better RFS (median: 19.5 vs. 10.5 months, p = 0.021) and OS (median: 32 months vs. 13 months, p = 0.055). This study was limited to a single university hospital, so the prognostic effect of immunophenotypic cluster should be further validated. Recently, IDH1, IDH2 and DNMT3A mutations have also been reported in AML patients with NPM1 mutation [32–34]. Correlating the immunophenotypic cluster with new biomarkers may also provide more insight into the molecular mechanisms of leukemogenesis in the future.