Changes in Type I Mutations Between Initial and Relapse Samples of Pediatric AML Patients Occur Frequently and Are Associated with Time to Relapse

Activating type I mutations provide cells with proliferative and survival advantages. Together with type II abnormalities, which cause a differentiation arrest and an increase in self renewal properties, they cooperate to cause acute myeloid leukemia (AML). Most studied type I mutations for their presumed prognostic relevance are the family of RAS oncogenes (e.g. N-RAS, K-RAS) and receptor tyrosine kinases (FLT3 and KIT). Relevant type II abnormalities include for example the NPM1 and CEBPα genes. We previously showed in paired initial and relapsed samples of 80 AML patients (40 children and 40 adults) and reported a shift from FLT3/ITD positive status at diagnosis to negative at relapse, associated with prolonged time to relapse, whereas a shift from FLT3/ITD negative to positive to be associated with a shorter time to relapse (Cloos et al., Leukemia, 2006). Here we extended the mutation analysis of paired initial and relapsed pediatric samples to study the stability of other mutations, besides FLT3. Samples of 34 pediatric AML patients were analyzed. Patients were treated with protocols of the BFM-AML Study Group or Dutch Childhood Oncology Group between 1992 and 2004. Using capillary gel electrophoresis based fragment analysis, we analyzed the patient samples for insertions/deletions in exons 14,15 and 20 of FLT3, exon 11 of KIT, exon 12 of NPM1 and 3 hotspots in the CEBPα gene. FRET based melting curve analysis or high resolution melting curve analysis were used to detect point mutations in exons 8, 9 and 17 of KIT, exons 3 and 13 of PTPN11, codon 12/13 and 61 mutations of N-RAS and codon 12/13 mutations of K-RAS. The frequencies of mutations and instabilities are summarized in Table 1. We found no mutations in CEBPα and PTPN11. Instabilities were found in 14 out of 34 (41%) patients (11/34 when excluding FLT3 mutations). Patients with and without a type I mutation at initial diagnosis had no significantly different mean time to relapse (17.2 (n=20) vs 18.6 (n=14) months, p=0.80). However, when patients are stratified according to the presence of a type I mutation at relapse (Table 2), independent of the presence at diagnosis, they had a significantly shorter mean time to relapse than patients without a type I mutation at relapse (8.3 vs 21.7 months, p=0.016). We defined poor prognosis mutation status as presence of type I mutation at relapse whereas a favorable prognosis as no type I mutation or presence of NPM1 mutation at relapse. Remarkably, patients who acquired a poor prognosis mutation status at relapse showed a similar short time to relapse as compared to those who had the mutation at diagnosis and retained the mutation at relapse (Table 2). Patients with an acquired favorable mutation status had a time to relapse comparable to those who already had a favorable prognosis at diagnosis. In conclusion, nearly 50% of pediatric AML patients show instability of type I mutations, with both gains and losses from diagnosis to relapse. This indicates that mutation profiling at diagnosis is not always sufficient for accurate risk assessment. Therefore, additional molecular characterization of AML at minimal residual disease and relapse might improve prognostics and guide targeted therapy. Possible explanations for mutational shifts are:

• expansion of minor leukemic sub-clones, present, but not detected at diagnosis;

• appearance of new leukemic sub-clones during therapy due to genomic instability. Future research in a larger patient group will focus on the detection and characterization of malignant sub-clones.

Table 1: Frequencies of mutations in genes of interest and their instabilities (N=34)

Table 2: Relation between mutation status at relapse and time to relapse