New horizons in the characterization and treatment of acute myeloid leukemia

Although the prognosis of pediatric AML patients has improved significantly over the last 30 years, 30-40% of patients relapse and 5-10% of patients die of the toxicity of treatment. Therefore, novel treatment approaches are needed. In the first part of this thesis (chapter 2-5), we described the poor prognosis of relapsed pediatric AML and we identified type I mutations in pediatric AML which might be used as targets for treatment. In the second part of this thesis (chapter 6-8), we examined the in vitro effects of targeted therapeutics and related sensitivity to these drugs to the expression and the presence of mutations of the specific targets. In chapter 2 we studied the clinical outcome of all pediatric AML patients initially diagnosed between 1980 and 1998 who subsequently relapsed (N=113). Most patients (63%) relapsed within one year after reaching first complete remission (CR1). In 80% of patients reinduction therapy was given with curative intent. No uniform treatment protocol was used. CR2 was achieved in 63% of patients, however the probability of 5 year overall survival (10-year pOS) was only 16%. In univariate analysis a short CR duration (<1 year) and FAB M4 were associated with poor survival, while patients with FAB M5 had a borderline significantly improved pOS. Stem cell transplantation (SCT) was performed in 25 patients after achieving CR2. In multivariate analysis, including SCT as a time-dependent variable, CR1 duration, FAB M4 and FAB M5, no factor was significantly associated with pOS. A significant proportion (24%) of children in CR2 who did not receive an allogeneic stem cell transplantation are long-term survivors. Half of the survivors suffered from late effects of treatment, especially the children who had been transplanted. This study shows that there are children with relapsed AML who can be cured without a SCT. It is important to identify these patients and spare them a SCT which is associated with frequent and severe long-term side-effects. The prognosis of relapsed AML is poor and collaborative studies with novel agents and new treatment schedules are necessary to improve outcome.
In the study described in chapter 3 we genotyped 150 pediatric AML samples for mutations in KIT (exons 8, 17), NRAS and KRAS (exons 1, 2) and FLT3/ITD.58 40% of children with AML had a mutation in KIT (11.3%), RAS (18%) or FLT3/ITD (11.1%).
We found interesting non-random associations between type I and type II mutations. Seventy percent of the core binding factor (CBF) leukemia cases had a mutation in KIT or RAS. Mutations in RAS (30%) or FLT3/ITD (20%) were frequently
found in association with a normal karyotype. Patients with a FLT3/ITD mutation had a significantly worse clinical outcome, but the presence of a KIT or RAS mutation did not significantly influence clinical outcome in our study. Small patient
numbers precluded a meaningful analysis of outcome within the CBF AML subgroup. We demonstrated that KIT exon 8 mutations result in constitutive ligandindependent kinase activation which could be inhibited by imatinib. Remarkable in
our study was the low frequency of KIT, RAS and FLT3 mutations in FAB M5 compared to non-FAB M5 AML (10% vs. 45%, p=0.011). It was previously reported that PTPN11 mutations were frequent in FAB M5 AML.203;204 We therefore
hypothesized that PTPN11 might be involved in the development of FAB M5 AML and investigated the prevalence of PTPN11 mutations in this patient cohort in chapter 4.79 We enriched this cohort with another 24 pediatric FAB M5 AML
patients, for a total of 55 FAB M5 AML cases. The PTPN11 mutation prevalence did not differ significantly between the FAB M5 and the non-FAB M5 cohorts (7.3% vs. 5.0%, p=0.73). Overall, we found a KIT, RAS, FLT3/ITD or PTPN11 mutation in 44% of patients. There were large differences between subgroups, as in CBF AML we found a KIT, RAS, FLT3/ITD or PTPN11 mutation in 70% of patients, while in FAB M5 this was only 27%. For some of these mutated proteins, novel drugs are now available which might be useful in the treatment of AML. In chapter 5 we determined FLT3 internal tandem duplications (FLT3/ITD) and D835 point mutations in paired initial and relapse samples from 80 pediatric and adult AML patients.80 Only one D835 point mutation was found, in an initial pediatric AML sample. A FLT3/ITD was present in 26% of newly diagnosed and 27% of relapsed AML samples. FLT3/ITD status changed between diagnosis and
relapse in 14 cases. In 4 patients the FLT3/ITD became undetectable at relapse, in 5 patients FLT3/ITDs were only detected at relapse, and in 5 patients the length or number of ITDs changed. FLT3/ITD positivity was related to a significantly shorter time to relapse, which was most pronounced when the ITD positive status was found at relapse. These results indicate that FLT3/ITD is not conserved between initial diagnosis and relapse and this makes it unsuitable as an MRD marker. In the second part of this thesis we focused on targeted treatment of AML. Novel drugs are necessary to improve the prognosis of children with AML preferably without additional toxicity. Specifically killing the leukemic cells with targeted drugs is an appealing concept as this might limit the side-effects of treatment. In chapter 6 we compared the cytotoxicity of free daunorubicin (DNR) and liposomal daunorubicin (L-DNR) in acute leukemia, normal bone marrow (N BM) and peripheral blood (PB) samples.138 There was strong cross-resistance between DNR and L-DNR, and within AML and ALL there were no significant differences between the sensitivity to DNR or L-DNR. Leukemic samples were significantly more sensitive to DNR and L-DNR than N BM and N PB mononuclear cells, reflecting the therapeutic index of these drugs. In conclusion, this study has shown that there is no significant difference in cytotoxicity between DNR and L-DNR and that therefore, L-DNR might be more effective clinically, as it is employed in higher dosages and has a greater AUC.