Cup-like acute myeloid leukemia: new disease or artificial phenomenon?

Abstract

We investigated cup-like nuclear morphology of acute myeloid leukemia blasts in 266 randomly selected patients and its association with hematologic findings, disease markers and outcome data. Cup-like acute myeloid leukemia was diagnosed in 55 patients (21%). It was associated with female sex, high white blood cell and blast cell counts, normal karyotype, and low CD34 and HLA-DR expression. Mutations of FLT3, NPM1 or both were detected in 84.9% compared with 58.1% in cases without this morphology (p=0.001). There was no influence on response to treatment or survival. Therefore, cup-like nuclear morphology is an indicator of normal karyotype and should guide more specific molecular analyses.

Introduction

There is an increasing role for molecular and cytogenetic findings in acute myelogenous leukemia (AML). In some categories, a correlation with morphologic features can be found, e.g. AML with inv(16), acute promyelocytic leukemia. More recently, several novel abnormalities have been identified, e.g. mutations of FLT3, CEBPA, NPM1 and RUNX1, some of which are predominantly found in patients with normal karyotype.1 Although an association with certain morphologic features has been described (e.g. FAB M5 morphology and FLT3 mutations), no highly specific morphology has been identified so far. Two recent studies2, 3 reported cases of AML showing a distinctive, cup-like nuclear indentation in myeloid blast cells. This was associated with a lack of CD34 expression, a higher frequency of normal cytogenetics, and mutations in NPM1 and FLT3 genes. Although a significantly higher rate of complete remissions was found in one of the studies, no impact on survival could be identified. We have previously reported on the prevalence and impact of mutations in FLT3 and/or NPM1 genes in patients with AML.4,5 A high mutant-to-wild-type (wt) ratio (>0.78) of the FLT3 internal tandem duplication (ITD) was an independent prognostic factor for shorter overall and disease-free survival. By contrast, single NPM1 mutations were associated with significantly better outcome. Because of this strong impact of FLT3 and NPM1 mutations on patients’ outcome, we re-evaluated the association of cup-like nuclear morphology with clinical and molecular features as well as outcome in AML patients.

Design and Methods

The data reported comes from the multicenter AML96 protocol of the Deutsche Studieninitiative Leukämie (DSIL). Detailed descriptions of this study have been previously reported.4–6 All patients received two cycles of standard induction chemotherapy. In patients under 60 years of age, postinduction treatment was stratified according to cytogenetic risk groups (low, intermediate, high risk) and included conventional chemotherapy (intermediate vs. high dose cytarabine), autologous or allogeneic stem cell transplantation. Patients with acute promyelocytic leukemia (AML FAB M3) were treated in another trial and therefore are not included in the present analysis. Complete remission (CR) was defined as the presence of less than 5% of blast cells in standardized bone marrow aspirate after the second induction cycle. The study was approved by the Institutional Review Board (IRB) of Dresden University Hospital.

All investigated samples were obtained at the time of diagnosis. Smears of peripheral blood (PB) and/or bone marrow (BM) from 266 randomly selected patients (male, n=148; female, n=118; median age 57 years) were re-evaluated for blast cell morphology. Two-hundred blast cells in PB and BM each were differentiated by two consecutive investigators (step 1) to determine the percentage of cells with cup-like nuclear invagination (spanning at least 25% of the nuclear diameter). The slides were then seen by two other investigators (step 2) who assigned the AML to a cup-like positive or a cup-like negative cohort without knowing the results from step 1. A comparison of the results from steps 1 and 2 was used to define cut-off (step 3). Both cohorts (cup-like positive vs. cup-like negative defined by cut-off value) were then analyzed for FAB-type, number of PB and BM blasts, CD34 and HLA-DR expression, karyotype abnormalities, molecular aberrations, as well as response and survival data. Antigen expression was detected by a standard indirect immunofluorescence method as previously described.7 Cytogenetic analyses were performed after routine preparation of fresh material using G-banding staining and fluorescence in situ hybridization (FISH). Molecular evaluation focused on mutations in FLT3 gene (internal tandem duplication, ITD; mutation in tyrosine kinase domain, TKD), NPM1 gene, and partial tandem duplication (PTD) mutations of the mixed lineage leukemia (MLL) gene. The techniques have recently been described in detail.4,5,8,9 In addition, electron microscopy was performed in cases with high numbers of cup-like blasts.

Figure 1.Typical morphology of myeloid blasts with cup-like nuclear invagination in AML. (A) The “cup” seems to be a part of the nucleus in Pappenheim staining, however it is positive for myeloperoxidase (MPO) in reactive blasts. (B) MPO-positive and MPO-negative cup-like blast cells in a patient with AML FAB M1.

Results and Discussion

The typical morphology of blasts with cup-like nuclei is demonstrated in Figure 1. Electron microscopy revealed that the morphologic phenomenon is caused by an accumulation of cytoplasmatic organelles rather than by primary nuclear changes (Figure 2, online supplement). The median percentage of blast cells with cup-like nuclei was higher in PB compared with BM (2.0% vs. 0.5%). In both cohorts, the percentage of cup-like cells correlated with the total number of blasts (PB, r=0.339; BM, r=0.263, p<0.01). Based on the results presented in Table 1 we used a cut-off of 5% to define cup-like positive AML (n=55) in the further analysis. FAB types M1 and M2 represented 62% of patients with cup-like AML. However, the frequency of cup-like positive cases among the monocytic (M4, M4Eo, M5a, M5b) and non-monocytic leukemias (M0, M1, M2, M6, M7, RAEB-T) was similar (20/89, 22.5% vs. 35/177, 19.8%; p=0.632). Table 2 summarizes the results of flow cytometric and genetic analyses. Cup-like nuclear morphology was associated with lower expression of CD34 as well as HLA-DR. However, this was due to highly significant differences among the non-monocytic leukemias, while an impact of nuclear morphology was not found to affect the expression of these antigens in the monocytic entities (data not shown). Cup-like morphology was significantly associated with normal karyotype, and molecular aberrations were also more often found in this cohort. In fact, in cup-like AML, FLT3 (ITD or TKD) mutations, NPM1 mutations or both were detected in 84.9% of cases, compared with 58.1% in the cup-like negative AML (p=0.001). We also analyzed the frequency of these mutations in monocytic leukemias (FAB M4, M4Eo, M5a, M5b) and non-monocytic entities (FAB M0, M1, M2, M6, M7, RAEB-T). In monocytic AML (with cup-like nuclear morphology), the frequency of FLT3 mutations, NPM1 mutations, or both were 30%, 25%, and 30% compared with 21.2%, 3.0%, and 60.6% in the non-monocytic leukemias. Mutations in FLT3 and NPM1 genes were significantly more frequent in cup-like AML compared with cases without this morphology among the non-monocytic FAB types (Table 1). By contrast, nuclear morphology had no impact on the frequency of these mutations among the monocytic leukemias (cup-like positive vs. cup-like negative: FLT3 mutations, 60% vs. 58%, p=1.000; NPM1 mutations, 55% vs. 46%, p=0.611). There was no statistical difference in the number of patients with MLL-PTD mutations between cup-like positive and negative diseases, either in the entire cohort or in the subgroups.

The rate of complete responses in cup-like positive and cup-like negative AML were 49.1% vs. 54.0% respectively (p=0.546) for the entire group and 40.0% vs. 53.5% respectively (p=0.187) in the non-monocytic entities. The median overall survival for the patients with cup-like positive and cup-like negative AML was 8.4 and 12.2 months (p=0.510). There was also no difference in median disease-free survival between the two cohorts of patients (p=0.739). In addition, we analyzed the subgroups of young patients <60 years, patients with FLT3 or NPM1 mutation, patients with FLT3 and NPM1 mutation, and patients with FLT3 ratio >0.8. However, cup-like nuclear morphology was not shown to have any impact on response rate or survival parameters.

Table 1.Number of cases with ≥2.5%, ≥5.0% and ≥10% cup-like blasts in PB and/or BM (step 1); results from identification of cup-like AML by subjective decision only (step 2, cases with divergent decision were discussed between the investigators); cut-off was defined on the basis of comparison of the results from steps 1 and 2.

At least for the experienced hematologist, cells with cup-like nuclear morphology are very easy to identify. Therefore we used a cut-off of ≥5% blast cells with typical morphology in blood and/or bone marrow for the definition of cup-like AML for the statistical analyses, which was lower than in the studies by Kussik et al.2 and Chen et al.3 The phenomenon seemed to be more pronounced in the peripheral blood and this was in line with the higher median percentage of cup-like positive blast cells in this group. Whether there really are more cup-like cells in the PB, or whether the phenomenon is induced by preparing the smears, remains hypothetical. Cellularity in the blood is very much lower than in the marrow and the specific morphology might therefore be easier to identify. A major finding from this series is that cup-like nuclear blast cell morphology is not restricted to acute myeloblastic leukemias as was suggested by other authors.2,3 In terms of HLA-DR expression, data are controversial. Kussik et al.2 reported a complete or partial loss of HLA-DR expression in cup-like AML. By contrast, cup-like AML was associated with a significantly increased expression of this antigen compared with a control group in the study by Chen et al.3 In our series, an association of nuclear morphology and HLA-DR was found only in the non-monocytic FAB types, not in the monocytic leukemias. The prognostic relevance of mutations in FLT3 and NPM1 genes has been proved in several trials.1 If not associated with FLT3 -ITD mutations, mutant NPM1 identifies a subgroup of patients with favourable prognosis, especially in AML with normal karyotype. By contrast, the negative prognostic impact of mutant FLT3 dominates in patients carrying both aberrations.5 In our study, about 70%, 60%, and 50% of cases with cup-like positive AML were positive for FLT3 -ITD mutation, NPM1 mutation, or both, respectively. Therefore, the detection of characteristics similar to those described for the FLT3 and NPM1 positive diseases is not surprising (high WBC and blast cell count at diagnosis, predominance in females and patients with normal karyotype, low expression of CD34 on blasts). Mutations of FLT3 and NPM1 are known to be more frequent in monocytic leukemias. However, a comparison of the frequency of these aberrations in patients with cup-like AML showed no difference between the monocytic and non-monocytic FAB types. Furthermore, the higher rate of FLT3 and/or NPM1 mutations in cup-like AML was mainly due to differences between cup-like-positive and -negative cases among the non-monocytic FAB-types, whereas the impact of nuclear morphology on the frequency of these mutations among myelomonocytic and monocytic leukemias was less pronounced. Therefore, cup-like nuclear morphology is indicative, but not specific, for the presence of potential risk-relevant mutations, particular in non-monocytic AML. Although a higher rate of complete remissions was found by Chen et al.,3 we could not show that cup-like morphology had any impact on response to treatment or survival either in the entire cohort or in subgroups. It may be that the number of patients even in our study is too small to detect differences in outcome. However, given the high frequency of molecular aberrations with very different prognostic consequences in cup-like AML (FLT3 and NPM1), it seems very unlikely that morphology should represent an independent predictive marker. Even more exciting is the question as to which biologic links exist between molecular aberrations and distinct nuclear morphology. A recently published model from HL-60 cells10 has shown that the changes in the nuclear shape during granulopoiesis involve factors that increase the flexibility of the nuclear envelope, augment connections to the underlying heterochromatin, and promote distortions imposed by the cytoskeleton. As demonstrated by ultra structural investigation, as well as by myeloperoxidase staining, the cup is filled in with cytoplasm and organelles. Due to the nucleocytoplasmatic shuttling properties of NPM1, a mutation in this gene can be considered as a causative event. While the bulk of physiological NPM1 protein resides in the nucleolus, mutant NPM1 protein is mainly detectable in the cytoplasm.5,11

Table 2.Comparison of patients with cup-like positive and cup-like negative AML in terms of epidemiologic, hematologic, cytogenetic and molecular parameters. The left of the table shows the results for the entire cohort including all FAB-types, whereas a second analysis (right of the table) was restricted to patients with non-monocytic leukemias (FAB M0, M1, M2, M6, M7, RAEB-T)

Altogether, our study supports data showing a close association between the distinctive cup-like nuclear morphology in AML blasts, specific clinical features and a high frequency of molecular aberrations with prognostic impact. This morphology can be, therefore, an indicator of normal karyotype and specific analyses of FLT3 and NPM1 mutations.

Footnotes

• The online version of this article contains a supplemental appendix.
• Authorship and Disclosures FK and US performed all the morphological investigations, RF and GB performed the ultrastructural study, FK, CT and UO wrote the manuscript, UO and RR performed the flow cytometric analyses, BM performed the cytogenetic analyses, SS was responsible for the statistics, MS and GE were responsible for the patient treatment and carefully checked the manuscript, and CT was the principal investigator who contributed the idea for the study and checked all the data. The authors reported no potential conflicts of interest.

• Received August 1, 2007.
• Accepted November 28, 2007.