NPM1 but not FLT3-ITD mutations predict early blast cell clearance and CR rate in patients with normal karyotype AML (NK-AML) or high-risk myelodysplastic syndrome (MDS)

In patients with a normal karyotype acute myeloid leukemia (NK-AML) submicroscopic genetic markers are essential for risk stratification regarding response to induction therapy and survival. NPM1 is the most frequent single genetic abnormality (46%-62%) and is associated with a favorable clinical outcome.^1,2  The presence of a FLT3-ITD mutation found in approximately 30% of patients^3,4  represents an unfavorable prognostic parameter for survival.

The positive prognostic effect of NPM1 mutations on response to induction therapy and long-term outcome was evident only in patients lacking the FLT3-ITD mutation. In addition to FLT3-ITD and NPM1, several other genetic markers^5-9  have prognostic impact in NK-AML. Kern et al¹⁰  showed that not only pretreatment characteristics but also early response parameters (early blast clearance) affect outcome. In these analyses, a positive prognostic impact of an adequate early blast clearance (< 10% residual blasts) on complete remission (CR) rate and long-term outcome was found. To gain further insight into the mechanisms mediating the favorable prognostic impact of NPM1 mutations, we investigated effects of the NPM1 mutation combined with or without the FLT3-ITD mutation on early response parameters.

This analysis was based on patients with newly diagnosed NK-AML treated within a prospective randomized multicenter trial according to the AMLCG 1999 study protocol. The details of the study have been published previously.¹¹  Cytomorphology, cytogenetic, and molecular analyses of bone marrow (BM) aspirates were performed according to standard protocols.^1,12,13 

All pretreatment clinical (age, sex, performance status [ECOG score], AML de novo, white blood cell count [WBC], hemoglobin level, platelet count, lactase dehydrogenase [LDH], BM blasts) and molecular markers (NPM1, FLT3-ITD, FLT3-TKD, CEBPA, and MLL-PTD) were included in univariate and multivariate analyses. Outcome parameters, CR rate, and early blast clearance were dichotomous. For univariate analyses, we performed cross tables and 2-sided exact fisher test for categoric characteristics and univariate logistic regression for categoric and continuous parameters. In multiple logistic regression, independent prognostic factors were identified by backward elimination using the Wald statistic^14,15  with significance level α = .05. Since with regard to overall survival an interaction effect of NPM1 and FLT3-ITD is known,^1,3  we analyzed a potential interaction effect on CR rate or early blast cell clearance by including the interaction term together with the main parameters NPM1 and FLT3-ITD in multiple logistic regression.

We investigated the influence of pretherapeutic markers on early therapeutic response parameters in patients with NK-AML enrolled in the AMLCG 1999 trial.¹¹  Patient characteristics are summarized in Table S1 (available on the Blood website; see the Supplemental Materials link at the top of the online article).

In 690 patients the FLT3-ITD mutation status and the NPM1 mutation status could be analyzed. The NPM1 mutation was present in 51% (n = 352/690) of patients. The FLT3-ITD mutation was found in 29% (n = 200/690) of patients.

Patients with known NPM1 and FLT3-ITD mutation status were divided into 4 groups: NPM1⁺/FLT3-ITD⁻ (n = 211/690, 31%), NPM1⁺/FLT3-ITD⁺ (n = 141/690, 20%), NPM1⁻/FLT3-ITD⁻ (n = 279/690, 40%), and NPM1⁻/FLT3-ITD⁺ (n = 59/690, 9%). In 598 patients with known NPM1/FLT3-ITD mutation status information about the early blast cell clearance was available.

Of 690 patients, 66% achieved a CR, 10% had persistent leukemia (PL), 15% died of early death (ED), and 9% remained hypoplastic. Seventy-five percent of patients carrying the NPM1 mutation and 57% of NPM1-negative patients achieved a CR (P < .001). CR rates did not differ significantly between the FLT3-ITD⁺ (68%) and the FLT3-ITD⁻ (65%) group (P = .480). Response to induction therapy differed according to the combination of the NPM1 and FLT3-ITD mutations. In NPM1 mutated AML, the CR rates were 77% (NPM1⁺/FLT3-ITD⁻) and 71% (NPM1⁺/FLT3-ITD⁺). Significantly lower CR rates of 56% (NPM1⁻/FLT3-ITD⁻) and 58% (NPM1⁻/FLT3-ITD⁺) were found in NPM1 unmutated AML (P < .001).

The amount of residual leukemic blasts in the BM measured 1 week after the end of induction therapy (ie, on day 12 in the HAM regimen; on day 16 in the TAD regimen) can be used as an early independent prognostic parameter of response to therapy⁹  and is referred to as early blast cell clearance. In our data set the early blast clearance did not differ significantly between patients treated with either TAD or HAM as first induction course (P = .661). Four-hundred nine (68%) of 598 patients showed less than 10% of BM blasts 1 week after the end of induction chemotherapy.

Eighty percent of NPM1⁺ patients (80%: NPM1⁺/FLT3-ITD⁻; 79%: NPM1⁺/FLT3-ITD⁺) but only 57% of patients without the NPM1 mutation (58%: NPM1⁻/FLT3-ITD⁻; 52%: NPM1⁻/FLT3-ITD⁺) showed a residual blast cell percentage of less than 10%. Thus, the NPM1 mutation was associated with a significantly higher blast clearance (P < .001) compared with the NPM1 wild-type (WT) genotype.

Seventy-one percent of FLT3-ITD–positive and 67% of FLT3-ITD–negative patients showed a blast reduction to less than 10%. Thus the presence of the FLT3-ITD had no influence (P = .498) on the blast cell clearance (Table 1).

Similar results were seen in younger (< 60 years) and older (≥ 60 years) patients (Table S2).

The CR rate was analyzed in a model of the 9 clinical and 5 molecular independent prognostic factors mentioned in “Methods.” Parameters with prognostic impact on the CR rate in univariate analyses included the presence of the NPM1 mutation (OR: 2.31; 95% CI: 1.67-3.19) and de novo AML (OR: 2.14; 1.44-3.17), as well as the presence of the MLL-PTD mutation (OR: 0.46; 0.26-0.82), age (OR: 0.78; 0.69-0.88), and WBC (OR: 0.75; 0.60-0.95). In multivariate analysis, the parameter with independent favorable impact on CR rate was the presence of the NPM1 mutation (OR: 2.81; 1.84-4.29), whereas a high WBC (OR: 0.53; 0.39-0.72) and higher age (OR: 0.81; 0.69-0.94) were significantly associated with a lower CR rate.

Regarding the effect on early blast clearance in univariate analyses, the presence of the NPM1 mutation (OR: 2.92; 2.03-4.20) and a high LDH level (OR: 1.97; 1.11-3.50) were strong positive indicators, whereas the MLL-PTD mutation (OR: 0.46; 0.24-0.89) negatively influenced blast clearance. Parameters not showing a significant effect included the hemoglobin level, WBC, platelets, BM blasts, age, sex, performance status, AML de novo, and mutations of FLT3 and CEBPA. In multivariate analysis, independent predictive variables were the NPM1 status (OR: 3.12; 2.01-4.83) and BM blasts at diagnosis (OR: 0.90; 0.82-0.98; Table 2).

Our data show that the presence of a NPM1 mutation is associated with the initial response to chemotherapy and thus represents a marker of sensitivity toward induction chemotherapy in vivo.

Pathophysiologically, this hypothesis is supported by in vitro experiments in which leukemic blasts carrying the NPM1 mutation¹⁶  showed significantly higher rates of apoptosis after treatment with chemotherapeutic agents compared with the NPM1 WT leukemic cells.¹⁷  Abnormal activation of the transcription factor NF-κB has been described in leukemic blasts resistant to chemotherapy.¹⁸  Cilloni et al¹⁷  showed that NPM1 plays a central role in enhancing apoptotic cell death in AML by interacting with NF-κB. Interestingly, NF-κB activity did not differ in NPM1⁺/FLT3-ITD⁺ and NPM1⁺/FLT3-ITD⁻ blasts,¹⁷  supporting our in vivo data that the presence of the NPM1 mutation is associated with a higher sensitivity toward induction chemotherapy regardless of the FLT3-ITD mutation status.

These data allow further insight into the biology of AML and provide a rationale for the better clinical outcome of this genetic AML subgroup.

This work was supported by Deutsche Forschungsgemeinschaft Grant SFB 684/A12 (to K.S.).

Contribution: F.S. performed statistical analysis and wrote the paper; E.H., M.U., S.F., A.H. and M.C.S. performed statistical analysis; S.S., A.D., T. Benthaus, G.M., and E.Z. performed molecular diagnostics; W.E.B. and B.J.W. coordinated the inclusion of patients into the AMLCG99 study and acquired patient data in cooperating hospitals; S.K.B., M.F.-B., C.B., J.B., and K.S. performed central diagnostics; K.S. wrote the paper; and W.E.B., T. Buechner, B.J.W., and W.H. were principal investigators of AMLCG99 study.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: K. Spiekermann, Department of Medicine III, University Hospital Grosshadern, Marchioninistr 15, 81377 Munich, Germany; e-mail: karsten.spiekermann@med.uni-muenchen.de.