Correlation of minimal residual disease cell frequency with molecular genotype in patients with acute myeloid leukemia

Abstract

Background About 70–80 percent of patients with acute myeloid leukemia enter complete remission, but at least half of these patients who achieve remission go on to relapse. Improved treatment is likely to come from increasing the time to relapse, especially for younger patients. With the vastly increasing number of targeted therapies there is a strong need for short-term end-points to efficiently test such therapies for further pursuance. Minimal residual disease assessment may offer such an end-point since it is a strong independent prognostic factor. As proof of principle we examined this concept for FLT3-ITD status at diagnosis.Design and Methods We determined FLT3-ITD status in bone marrow samples from 196 patients with newly diagnosed acute myeloid leukemia. The frequencies of residual leukemic cells of these 196 patients were assessed in 267 follow-up bone marrow samples using immunophenotypic assessment of minimal residual disease.Results The median frequency of residual leukemic cells after the first cycle of chemotherapy was 8.5-fold higher in patients with FLT3-ITD than in those with wild type FLT3. Such a difference translates into differences in survival, even if other potentially outcome-modulating mutations, such as NPM1, KIT, NRAS, KRAS, FLT3-exon 20 and PTPN11 are included in the analysis.Conclusions This study shows that it could be possible to study the efficacy of FLT3 inhibitors using the level of minimal residual disease as a short-term end-point.

Introduction

At present treatment of acute myeloid leukemia (AML) in adults is based on the assessment at diagnosis of a limited number of prognostic factors including cytogenetics. From a genetic point of view AML originates from a wide variety of acquired somatic mutations. The development of AML is now considered a two-step process: class 2 mutations such as t(8;21), AML1-ETO and inv(16) must work in concert with the so-called class 1 mutations. Examples of class 1 mutations are the activating mutations of receptor tyrosine kinases (KIT and FLT3), and mutations in the proto-oncogene RAS family members, NRAS and KRAS. These insights are being translated into the development of molecules that target these specific mutations directly or that act downstream in their disrupted signal transduction pathways. Numerous clinical trials using small molecule inhibitors, such as FLT3 inhibitors, in addition to conventional chemotherapy have been initiated.1,2

In order to have an effecient clinical trial design for the large variety of modulating agents and their combinations with other drugs, including classical chemotherapeutic agents, early efficacy read-out parameters are necessary. Until now, the achievement of complete remission, remission duration and overall survival have been used for this purpose. However, in the light of clone selection, shifts of (stem) cell subpopulations and multiple hit models, an earlier read-out is desirable.

The outgrowth of minimal residual disease (MRD) cells has previously been indicated to be responsible for the emergence of relapse.3 Moreover, the frequencies of MRD cells in bone marrow, characterized immunophenotypically by an aberrant phenotype, were shown to have a prognostic impact after induction and intensification therapy.4–7

We hypothesized that the frequencies of MRD cells after the first cycle of chemotherapy could be used as an early read-out parameter in future clinical trials using FLT3 inhibitors. To test this hypothesis, we studied the relationship between the presence of FLT3-internal tandem duplications (ITD) and MRD cell frequency. The presence of class 1 mutations other than ITD in exon 14 of the FLT3 gene (NPM1, KIT, NRAS, KRAS, FLT3-exon 20 and PTPN11) was accounted for in this analysis.

Design and Methods

Patients’ samples

The bone marrow aspirates, peripheral blood samples and clinical data presented in this study were obtained from 288 AML patients, treated at the Hematology Department of the VU University Medical Center (Amsterdam, The Netherlands). Informed consent was obtained from all patients, with approval of the institutional review board. The diagnosis of AML was established on the bases of morphology and immunophenotype, according to the French-American-British (FAB) classification. Cytogenetic aberrations were scored according to Grimwade et al.8

Treatment characteristics and definitions

Patients were treated according to the Dutch-Belgian-Swiss (HOVON) protocols (www.hovon.nl). Patients 60 years of age or younger were treated according to the HOVON 29 (during 1998–2000) or HOVON 42 (during 2001–2006) protocol. Patients over 60 years of age were treated according to the HOVON 32 (during 1996–2000) or HOVON 43 (during 2000–2006) protocol.

All follow-up samples were obtained after full hematologic recovery of the bone marrow. Hematologic recovery was defined as the time point at which the white blood cell count (WBC) was greater than 1.0×10/L, the granulocyte count was greater than 0.5×10/L and the platelet count was greater than 100×10/L. Complete remission was determined morphologically and defined by the presence of less than 5% blasts in the bone marrow, combined with recovery of the peripheral blood cell counts. Early/toxic deaths were defined as all deaths occurring within 7 days after completion of the first induction cycle or death during therapy-induced bone marrow hypoplasia. Overall survival was defined as the time interval between inclusion in the study and death, relapse-free survival as the time interval between achievement of complete remission and relapse, and disease-free survival as the time interval between diagnosis and the first event (relapse or death, whether or not disease-related) for patients who had achieved complete remission.

Establishment of leukemia-associated phenotypes

MRD cell frequencies were determined using leukemia-associated phenotypes (LAP). LAP were established on the blasts at diagnosis applying a double step immunophenotypic labeling procedure using fluorescein isothiocyanate (FITC)-, phycoerythrin (PE)-, peridin chlorophyll protein (PerCP)- and allophycocyanin (APC)-conjugated monoclonal antibodies, as previously described.7,9,10

In normal bone marrow these LAP are, by definition, absent or present in very low frequencies.11 Since the penetrance of LAP positivity on leukemic blasts at diagnosis is most often not 100%, the actual LAP frequency at diagnosis has to be taken into account when the frequency of LAP-positive leukemic blasts in follow-up samples is calculated. Corresponding isotype controls were included to test the specificity of staining for all markers. In the present cohort every patient displayed one or more LAP, thus enabling MRD detection during follow-up.

Minimal residual disease study group and minimal residual disease cell frequency

Since the association between MRD cell frequencies and FLT3-ITD status was the main focus of this study, we restricted the description of clinical characteristics to the 196 patients displaying an aberrant blast immunophenotype at diagnosis, which enabled MRD assessment in follow-up samples. Details are shown in the second column of Table 1. Overall 157 out of these 196 patients achieved complete remission; 128 after the first cycle of consolidation chemotherapy, 23 after the second cycle and one after the third cycle. Complete remission status in relation to chemotherapy cycle was unknown for three patients. Sixteen patients did not achieve complete remission after receiving two cycles of induction chemotherapy. Twenty patients experienced early/toxic death. Table 1 presents all these characteristics, subspecified for the patients included after the different MRD assessment times e.g. after first and second induction chemotherapy and after consolidation treatment. At diagnosis 179 bone marrow and 17 peripheral blood samples were obtained from the 196 patients. In patients whose peripheral blood was used for LAP assessment, the blast percentage was on average 57% (range, 4.6–94%). During follow-up 388 bone marrow samples were obtained at different time points, 175 after the first cycle of chemotherapy, 120 after the second cycle and 93 after the third cycle.

Table 1.Patients’ characteristics.

The total MRD study cohort thus included 196 patients, of whom 157 (80%) achieved morphological complete remission (Table 1). By definition the term MRD cell frequency applies to those patients who achieved complete remission (n=157). Based on this selection, MRD cell frequencies were determined after the first, second and third cycles of chemotherapy in 107/175, 95/120 and 60/93 samples, respectively.

Sample preparation

Genomic DNA or RNA was isolated from quickly frozen cell pellets obtained at diagnosis from all 288 patients included in the study. Genomic DNA was extracted using a QIAamp blood DNA extraction kit (Qiagen, Crawley, UK). Total RNA was extracted using RNABee solution (Tel-test Inc., Friendswood, TX, USA) according to the manufacturer’s recommendations.

Polymerase chain reaction (PCR) primers and conditions for FLT3, NPM1, KIT, NRAS, KRAS, FLT3-exon 20 and PTPN11 mutation analysis have been described12–14 previously and are available on request.

Statistics

MRD cell frequency was assessed both as a continuous variable and, by applying cut-offs, also as a dichotomous variable. The cut-offs applied were the median MRD value, and various others decided on the basis of optimized hazard rates (HR). Kaplan-Meier survival curves were constructed for subgroups of patients based on MRD cut-offs. The present study included patients with an immunophenotypically defined (using CD45 blast cell gating15) MRD cell frequency of less than 5% blasts (Table 1). In two cases this contrasted with the morphologically defined status of complete remission.

The relationship between MRD cell frequency and overall, relapse-free and disease-free survival and also between FLT3 status and these survival parameters were estimated for patients who were included in the MRD analysis. A Cox proportional hazard model was used for multivariate analysis of prognostic factors. Clinical pathological factors assessed in the multivariate analysis included FAB-class (categorical variable), cytogenetics (categorical variable) and WBC at diagnosis (expressed as a logarithm and applied as a continuous variable). Treatment protocol was included as a dichotomous variable: two protocols for younger patients (under 60 years old) and two protocols for elderly patients (60 years old or over). Age-related treatment protocols are dealt with separately since they had a different design. Type of consolidation treatment was assessed as a categorical variable, including four categories; third cycle of chemotherapy, allogeneic transplantation, autologous transplantation and no consolidation treatment.

The distribution of the clinical variables was compared using χ analysis for categorical variables and the non-parametric Mann-Whitney U test for continuous variables. Multivariate analysis of the impact of multiple mutations on MRD cell frequencies was conducted using either linear regression (in the case that MRD was studied as a continuous variable) or logistic regression (when MRD cell frequency was studied as a dichotomous variable). Patients still in remission were censored at the time of last follow-up.

The reported p values are two-sided and p<0.05 was considered statistically significant in all analyses. Calculations were performed using the SPSS software package version 11.0.1 (SPSS, Chicago, IL, USA).

Results

FLT3-ITD status and survival

In total 265 samples from the 288 newly diagnosed AML patients were successfully analyzed for ITD in exon 14 of the FLT3 gene. FLT3-ITD were found in 61/265 patients (23%). Failure of mutation analysis (23 patients) was due to either shortage of cells or low DNA yields. The percentage of FLT3-ITD in the total population (n=265) was comparable to the percentages observed in the subgroups of patients in the total MRD group (n=196) included in MRD assessment studies after the first (n=107), second (n=95) and third (n=60) cycles of chemotherapy, in which ITD were observed in 20.7%, 16.7%, 23.1% and 18.6% of the patients, respectively.

Assessment of the distribution of clinical characteristics across patients with FLT3-ITD (n=36) and those with wild type (wt) FLT3 (n=138) in the MRD study revealed an unequal distribution between the two groups for log WBC (higher in the FLT3-ITD group, Table 1), favorable cytogenetics (lower in the FLT3-ITD group, Table 1) and intermediate cytogenetics (higher in the FLT3-ITD group, Table 1). Cases with unknown data were excluded from the calculations. No significant differences were found for most FAB types, age-related treatment protocol or type of consolidation regimen. Trends were visible for percentages of patients diagnosed as having refractory anemia with excess blasts in transformation or AML secondary to a myelodysplastic syndrome (13.0% in the FLT3-wt group compared to 3.0% in the FLT3-ITD group) and FAB class M5 (present in 30.3% of patients with FLT3-ITD compared to in 19.8% with FLT3-wt).

In the total MRD population of patients (n=196) in whom FLT3-ITD status was evaluated (n=171), a FLT3-ITD was associated with shorter overall survival (HR 1.7; 95%CI 1.06–2.72; p=0.027), relapse-free survival (HR 2.1; 95%CI 1.16–3.62; p=0.013) and disease-free survival (HR 1.9; 95%CI 1.13–3.21; p=0.015). The adverse prognostic impact of an ITD in the FLT3 gene on overall, relapse-free and disease-free survival was maintained although the statistical significance was weaker because of the smaller numbers of patients in the subpopulations studied after the different courses of chemotherapy (HR after the first cycle 1.6/1.9/1.9; p=0.09/0.05/0.03; HR after the second cycle 1.7/2.0/1.9; p=0.11/0.05/0.05, HR after the third cycle 1.7/1.6/1.9; p=0.13/0.14/0.05).

Minimal residual disease cell frequency and survival

In agreement with our previous experience, MRD cell frequency in samples from 107 patients after the first cycle of induction chemotherapy predicted overall survival (HR 1.2; 95%CI 1.09–1.32; p<0.0001), relapse-free survival (HR 1.6; 95%CI 1.15–2.31; p=0.0006) and disease-free survival (HR 1.2; 1.08–1.34; p=0.001). The optimal cut-off in MRD cell frequency of 1.0% in the present cohort of patients revealed 3.7-fold and 4.1-fold higher risks of shorter overall survival (95%CI 1.63–8.30; p=0.002) and relapse-free survuval (95%CI 1.50–10.94; p=0.006), respectively, for patients with MRD cell frequencies of greater than 1.0%. The optimal cutoff of 1% remained significant in the subgroup of younger patients (p=0.02) and of borderline significance in the subgroup of elderly patients (p=0.08).

Based on the assessment of 96 patients after the second cycle of chemotherapy, MRD cell frequency, considered as a continuous variable, was also highly predictive of the patients’ overall, relapse-free and disease-free survival (HR 1.23; 95%CI 1.059–1.431; p=0.007, HR 1.20; 1.035–1.392; p=0.016 and HR 1.17; 95%CI 1.033–1.340; p=0.014), respectively. The optimal cut-off 0.8% remained significant in both the younger and older groups of patients (p<0.0001 and p=0.024, respectively).

Among the 60 patients who were assessed after the third cycle of chemotherapy, a higher MRD cell frequency also correlated with poorer overall survival (HR 1.42; 95%CI 1.11–1.81; p=0.005), relapse-free survival (HR 1.36; 1.08–1.71; p=0.010) and disease-free survival (HR1.43; 95%CI 1.13–1.81; p=0.003). The cut-off in MRD cell frequency of 0.4% resulted in the most significant difference with a 3.6-fold higher relative risk of relapse for the patients in the high frequency group. This optimal cut-off of 0.4% remained significant for the younger patients (p=0.01) although not for the elderly patients (p=0.31) as could be expected on the basis of the low number of subjects in this group (n=11).

FLT3-ITD status and minimal residual disease cell frequency

FLT3-ITD status was not associated with remission rate (p=0.55). In the FLT3-ITD group 28/30 (Table 1, column 4) achieved complete remission, while in the FLT3-wt group 114/126 (Table 1, column 3) did so.

However, in patients entering complete remission, FLT3-ITD status was strongly associated with MRD cell frequency after the first cycle of chemotherapy (n=96, p=0.002). After the second course of chemotherapy the presence of an FLT3-ITD remained associated with higher MRD cell frequency although this association was much less pronounced (n=91, p=0.082). After the third cycle of chemotherapy the association was lost (n=59, p=0.69).

Median MRD cell frequencies in the total population of patients dropped during consecutive chemotherapy cycles (median MRD cell frequencies after the first, second and third cycles were 0.06, 0.04 and 0.03, respectively). Since 0.01% is the lowest detectable frequency of MRD cells for immunophenotypic assessment of MRD,5,6 this reduces the differences in MRD cell frequencies between FLT3-wt and FLT3-ITD patients. It is likely that MRD cell frequencies would decrease further after additional cycles of chemotherapy.

Survival effects of minimal residual disease cell frequency in FLT3ITD and-FLT3-wt patients

The effect of MRD frequencies and their cut-offs on overall survival and relapse-free survival for patients with FLT3-ITD and FLT3-wt are displayed in Figure 1.

The distribution of MRD cell frequencies in FLT3-ITD and FLT3-wt samples after different cycles of chemotherapy is shown in Figure 2.

The results obtained for MRD after the first course of chemotherapy were analyzed in more detail: the median MRD cell frequencies in the two groups differed by a factor of 8.5: MRD% 0.34 (FLT3-ITD) versus 0.04 (FLT3-wt). Significantly more patients with FLT3-wt were found below the median cut-off for the FLT3-ITD group of 0.34% compared to above the cutoff (41/70 patients; χ test 4.401). Similarly, significantly more FLT3-ITD cases were found above the median cut-off for the FLT3-wt patients of 0.04% (15/19 patients; χ test 4.436). The 8.5-fold difference in MRD% translated into differences in median survival: among the total population of patients, the median overall and relapse-free survival of patients with MRD% of 0.01 (the lowest detectable MRD%) was, respectively 10.7 and 11.1 months, while the median overall and relapse-free survival of patients with MRD% that were 8.5-fold higher than the lowest detectable MRD% (range, 0.085–0.1) was 8.8 and 6.9 months, respectively.

This also held true for other 8.5-fold differences in ranges of MRD cell frequencies. This implies that the use of an inhibitor of tyrosine kinase activity could not only cause large differences in MRD cell frequency, but also differences in survival.

Figure 1.Relapse-free survival (RFS) and overall survival (OS) in FLT3-ITD and FLT3-wt patients based on chemotherapy cycle-specific cut-offs in MRD cell frequency. For each cycle of chemotherapy in the whole group of patients (FLT3-ITD + FLT3-wt) a cut-off in MRD frequency was determined that was best at dividing patients according to their survival (1.0 after the first cycle, 0.8 after the second cycle and 0.4 after the third cycle). Kaplan-Meier curves were constructed from these. Effects of chemotherapy cycle specific cut-offs in MRD frequency are shown for OS and RFS in patients with FLT3-ITD and FLT3-wt. Both FLT3-wt and FLT3-ITD patients with MRD frequencies higher than the cycle-specific cut-off had significantly or borderline significantly shorter OS and RFS compared to those patients with MRD frequencies lower than the cycle-specific cut-off. Black straight line: FLT3-ITD patients with MRD frequencies above the cut-off, black dotted line: FLT3-ITD patients with MRD frequencies below the cut-off, red straight line: FLT3-wt patients with MRD frequencies above the cut-off, red dotted line: FLT3-wt patients with MRD frequencies below the cut-off.

KIT, NRAS, KRAS, PTPN11, NPM1, FLT3 exon 20 mutations and FLT3-ITD status

Among other fractors, a poor response to therapy in part of the FLT3-wt group, may result from other mutations. Although none of the investigated mutations besides FLT3-ITD had a significant impact on overall, relapse-free or disease-free survival on individual bases (data not shown), a combined regression analysis showed that both KRAS and PTPN11 improved the overall prediction of the FLT3-status with respect to overall survival (p=0.03 χ 7.04 and p=0.03 χ 7.32, respectively). In the case of KRAS this contribution was possibly associated with the achievement of complete remission in the FLT3-wt patients, whereas PTPN11 was found to be associated with early death (Table 2).

Figure 2.Distribution of MRD frequencies in FLT3-ITD and FLT3-wt patients. Dot plots show the distribution of MRD cell frequencies (y-axis) in FLT3-ITD (solid black squares) and FLT3-wt patients (open black squares) in morphological complete remission after the first, second and third cycles of chemotherapy (x-axis). The median MRD frequencies in FLT3-ITD and FLT3-wt patients after the first cycle of chemotherapy differed by a factor of 8.5: 0.34% in the FLT3-ITD patients versus 0.04% in the FLT3-wt patients. For the second and third cycles the corresponding figures are 0.08% versus 0.03% and 0.02% versus 0.04%, respectively. These differences in median MRD frequencies between FLT3-ITD and FLT3-wt patients were significant after the first and the second cycles of chemotherapy (p=0.05). Median MRD frequencies for FLT3-ITD patients after the second cycle dropped by a factor of 4.25 compared to the first cycle and after the third cycle by a factor of 4.0 compared to the second cycle. The corresponding figures for FLT3-wt patients were 1.3 and 0.75, respectively.

NPM1 mutations were more frequently observed among the patients with FLT3-ITD (Table 2). In agreement with the favorable prognosis of NPM1 mutations in the FLT3-wt group,16 we observed a lower MRD percentage in FLT3-wt patients with NPM1 mutations than in FLT3-wt patients without NPM1 mutations (0.015 and 0.06, respectively). In FLT3-ITD patients, the presence of NPM1 mutations was associated with a higher MRD percentage; 1.20 versus 0.18 in NPM1-wt patients). In the total population of patients assessed after the first cycle of chemotherapy, the presence of an NPM1 mutation was associated with a lower median MRD percentage (0.035 versus 0.07 in NPM1-wt patients).

Discussion

In this study we show that the presence of FLT3-ITD at diagnosis of AML is associated with an 8.5-fold higher frequency of MRD cells after the first course of chemotherapy compared to that found in FLT3-wt patients. This difference translated into differences in overall survival, relapse-free survival and disease-free survival. Partial or complete inhibition of the aberrations resulting from the ITD would, therefore, translate into significant survival differences. Since the effects were very pronounced after the first cycle of chemotherapy, the MRD cell percentage at this time point offers a fast read-out end-point to assess the efficacy of FLT3 inhibitors in terms of survival.

From a theoretical point of view a higher MRD cell frequency in FLT3-ITD patients after the first cycle of chemotherapy should result in higher MRD cell frequencies after the second and third cycles of chemotherapy, since MRD cell frequency after all cycles of chemotherapy strongly predicts survival.5,6 The loss of association between MRD cell frequency and FLT3-status during consecutive cycles of chemotherapy most likely results from flow cytometric MRD detection limit (0.01%): in many cases this detection limit has already been reached after the second and third cycles, not allowing further discrimination between patients. This loss in ability of discrimination is not insuperable; however, the first cycle of chemotherapy offers the earliest and, thereby, most favorable point in time for the assessment.

Since the association between MRD cell frequency and treatment efficacy is not limited to flow cytometry, MRD cell frequencies after the second and third cycles of chemotherapy may function as proper read-outs for trials of targeted inhibitors in AML in the future, provided that sensitive molecular techniques are available for the majority of patients.

In general, each course of chemotherapy reduces the amount of MRD in patients as a whole:5,6 in the present study median MRD cell frequencies were 0.06, 0.04 and 0.03 after the first, second and third cycles of chemotherapy, respectively. However, Figure 2 shows the strong impact of the second and also the third cycles for the subgroup of patients with FLT3-ITD with 4.25 and 4-fold further reductions (total 17-fold) after the two extra courses. In contrast, among the group of patients with FLT3-wt, no reduction in median MRD cell frequency was seen after two extra courses of chemotherapy. From the large difference in MRD cell frequencies between FLT3-ITD and FLT3-wt patients it can be concluded that patients with a FLT3-ITD at diagnosis had a large benefit from the subsequent cycles of therapy.

For final proof of the utility of MRD cell frequency as a short-term read-out for the efficacy of targeted treatments, a large prospective study is needed. One may ask whether or not FLT3-wt patients should be included in the treatment group in such a study, since inhibitors may also exert effects in such patients as a result of non-specific effects. Clinical studies that randomize all patients between the use of an inhibitor versus no inhibitor would enable effectiveness of the inhibiors to be assessed both in patients with FLT3-wt and FLT3-ITD.

One of the major challenges and ultimate goals for the application of existing and new therapies is to be able to predict the response and survival of individual patients. For the area of overlap between MRD cell frequencies in wild type and ITD cases (Figure 2) this translates into two questions: (i) what is the cause of poor responses in some patients FLT3-wt¿ and (ii) why do some FLT3-ITD patients perform quite well¿ With regard to the first question, an intriguing possibility is that, irrespective of the cellular characteristics in part of any group of AML patients, pharmacokinetic resistance appears to overrule cellular resistance.10 With regards to the present study, this means that in the wild type group, but also in the ITD group, high levels of MRD may have resulted, in part, from pharmacokinetic resistance. Furthermore, we and others have recently shown that poor prognosis is not only a characteristic of patients with FLT3-ITD at diagnosis, but may also result from apparent post-diagnosis gains of FLT3-ITD in patients who had the wild type gene at diagnosis.12,17 The second question is more difficult to answer. Part of the solution might be post-diagnosis losses of FLT3-ITD, since we and others have shown that such cases are associated with a better prognosis.12,17 Whether such clone shifts can already be manifested during the first cycle of therapy is presently under investigation.

Table 2.Distribution of mutations specified for complete remission and FLT3-ITD status in patients after the first cycle of chemotherapy.

As common events take place in transduction and transcription pathways that are associated with malignant transformation of hematopoietic stem cells, levels of MRD cells may also be indicative of the efficacy of inhibitors that target mutations other than FLT3-ITD. Bearing in mind that MRD cell frequency is a very strong independent prognostic factor, it is likely that the present approach may also be applicable to other mutations. In fact, so far we have shown that other factors at diagnosis that have prognostic impact in terms of survival, i.e. apoptosis-related protein expression,18 P-glycoprotein activity19 and stem cell frequency,20 translate into differences in MRD cell frequency. These factors may explain why some FLT3-wt patients have a poor survival. For the FLT3-ITD patients the additional effects of other mutations are less pronounced since most of the additional mutations are seen in the FLT3-wt patients. Furthermore, the additional mutations are more or less scattered over the FLT3-ITD group, irrespective of MRD cell frequency.

Overall, this study reveals excellent correlations between FLT3-ITD status, MRD frequency and patients’ survival, enabling early evaluation of FLT3 inhibitor efficacy in a clinical setting.

Footnotes

• Funding: this manuscript was supported in part by funding from the Leukemia and Lymphoma Society (MCH) and a VA merit Review Grant (MCH).
• Authorship and Disclosures CJH, GJS, GJO and QW designed the research and wrote the manuscript; FD, CJH and PAM analyzed the data; MCH and AH performed the mutation analyses; AK and NF designed and performed the MRD cell frequency analysis; JB assisted in the statistical analysis. The authors reported no potential conflicts of interest.

• Received March 25, 2008.
• Revision received September 3, 2008.
• Accepted September 23, 2008.