We have previously reported that FLT3 internal tandem duplications (FLT3/ITD) were detectable in 11.5% of 234 pediatric AML samples at initial diagnoses, and conferred a poor prognosis (Zwaan et al, Blood 2003). However, no data were available on relapsed pediatric AML and on paired initial-relapse samples. Currently we have studied genomic DNA of 99 relapsed and 42 paired initial-relapse pediatric AML samples for FLT3/ITD. Samples were obtained from the AML-BFM-SG, the Dutch Childhood Oncology Group, and by centers participating in the International BFM Relapsed AML 2001/01 study.Of the 99 relapsed patients the median age was 10.2 years and 71% were boys. Median time from diagnosis to first relapse was 14 months. Patient characteristics of the 42 pairs were at diagnosis: median age 7.7 years, sex 80% boys; at relapse: median age 14.3 years. Median time from diagnosis to first relapse was 14 months. ITDs were detected by PCR amplification including a fluorescent labeled primer. Fragments were separated by capillary electrophoresis on an ABI 3100 and analyzed using Genescan software. Of the 99 relapsed samples, 20 were ITD positive (20.2%), with ITDs varying in length from 18–129 bp. ITDs occurred in 15% of children <10 years of age, and in 21% ≥ 10 years (p=0.5). ITD pos. patients were boys in 84% vs. 68% in wild type (wt) patients (p=0.16). WBC, blast percentage and FAB-types were not different between ITD and wt patients. The time-interval between diagnosis and relapse was significantly shorter in ITD pos. patients (10 months) than in wt ones (15 months, p=0.003).Flt3/ITD in paired initial vs relapse samples of pediatric AMLRelapse ITD posRelapse wild typeTotalInitial ITD pos6 (14%)4 (10%)10 (24%)Initial wild type4 (10%)28 (66%)32 (76%)Total10 (24%)32 (76%)42 (100%)Two of the 42 paired samples had 2 different length mutations at initial diagnosis and only 1 at relapse. Four acquired an ITD at relapse (one at 2nd relapse), which varied from 21–129 bp. FAB-types were: 1xM1, 2xM2, 1x M6. Median time to first relapse was only 6 months. Four patients (FAB-types were 2xM1 and 2xM2) lost the ITD, which at diagnosis varied from 30–72 bp and interestingly the median time to relapse was relatively long (median of 30 months). All ITDs were confirmed by sequencing and they were all located in exon 14. We conclude that: a) detectable FLT3/ITD is more frequent in the relapsed patients (±20%) compared to our previous study with patients at initial diagnosis (11.5%); b) patients with initially detectable FLT3/ITD relapse earlier than wt patients; c) there were no significant differences in clinical and cell-biological characteristics between wt and ITD patients at relapse d) gain and loss of ITDs at relapse does occur e) patients who gain an ITD seem to relapse early, patients who lose an ITD relapse relatively late. The changes in ITD status may reflect true de novo mutations or a selection of specific clones. In the latter case a small ITD positive subclone was not detected. We will further elucidate this with specific PCRs we are currently developing. These results have relevance for Flt3-targeted therapy and minimal disease monitoring.