Development of An Oligonucleotide Resequencing Array for Rapid Mutation Analysis in Acute Myeloid Leukemia with Normal Karyotype.

Abstract 705

During the last years AML with normal karyotype (NK-AML) has been gradually elucidated by molecular genetic mutations. The constellation of specific mutation patterns was shown to have prognostic relevance (e.g. NPM1 mutated/FLT3 unmutated correlate with good prognosis). The growing prognostic relevance and the development of specific drugs forces the need for tools that costsaving analyze a large number of genes within a short turn around time. Oligonucleotide resequencing arrays can explore base exchanges by standard design or known mutations by specific probe designs representing the exact sequence of the mutation. For analysis of NK-AML we have developed a customized oligonucleotide resequencing array (Affymetrix, Santa Clara, USA) covering 13 AML relevant genes (CEBPA, FLT3, JAK2, KIT, KRAS, MLL, MPL, NPM1, NRAS, PTPN11, RUNX1, TP53 and WT1). Special probe designs were used for the most frequent mutations like NPM1_A, NPM1_B, NPM1_D, JAK2V617F, KITD816V, FLT3D836, NRAScodon12, 13, and 61 mutations. 2 μg of DNA for each patient were required and the workflow comprises 128 PCR reactions, pooling, fragmentation, hybidization, staining, washing, scanning, and evaluation. All steps were optimized to allow completion within 3 days. For each patient the whole coding region of all genes (34 kb) was amplified in 128 PCR reactions. In total 64 NK-AML patients were analyzed that were precharacterized by standard methods for CEBPA, FLT3-ITD, FLT3-TKD, JAK2, KITD816, MLL-PTD, NPM1, NRAS, RUNX1 and WT1exon 7 and 9.The patients were selected based on the detection of at least 2 different mutations and these included CEBPA: n=16, FLT3-ITD: n=29, FLT3-TKD: n=11, JAK2: n=2, KITD816: n=4, MLL-PTD: n=11, NPM1type A: n=15, NPM1 type B: n=3, NPM1 type D: n=2, rare NPM1 types: n=8, NRAS: n=8, and RUNX1: n=12. Using this newly designed resequencing array and after evaluation with the JSI software (Medical System Kippenheim, Germany) identification of the previously known base exchanges was achieved in 11/11 in FLT3-TKD, 2/2 JAK2V617F, 8/8 NRAS, 3/4 KITD816V, 2/2 CEBPA and 2/2 in RUNX1 resulting in a sensitivity of 27/28 (96.4%). One KITD816 mutation was not detected as it was present in less than 10% of cells and thus was below the specifications of the array. In addition, by use of the mutation specific designs all NPM1 type A (n=15), B (n=3) and D (n=2) mutations and 7/8 of the rare NPM1 mutations were detectable, resulting in an NPM1 total detection rate of 27/28 (96.4%). As expected, with currently available software tools it was not possible to detect mutations that cause dose effects like FLT3-ITD or MLL-PTD or insertion/deletion mutations. In contrast, new mutations were detected in genes and regions that were not covered by standard techniques. These latter mutations were subsequently verified by Sanger sequencing: five so far undetected mutations in FLT3 were V491L; G549A, G481E, V194M, and N676T. Although the clinical relevance of most of these mutations is unclear N676T has been implicated in PKC412 resistance. Furthermore, four previously not described mutations were detected in JAK2 (V446G, G571S, E846D, R1063H).Three new mutations were found in MPL (W416X, W632C, L594W). Similar MPL mutations have been described in familial thrombocythemia and although these patients have a high risk for development of AML these mutations have so far not been implicated in de novo AML. Three new mutations were detected in WT1 (G44V, E47V, H137R), 3 in KRAS (Q61L (n=2) and V114L), one in PTPN11 (E76K) and four mutations of unknown significance in MLL (G276G, E502K, A822V, S901R). Thus, in total 24 new mutations were detected in 64 patients (38%) in addition to the respective two mutations detected by conventional techniques indicating that a significant subset of AML carry more mutations than reflected by standard approaches. Although the clinical significance of most of these mutations is unclear this may have implications in new drug development and study design. In conclusion, using this newly designed customized AML resequencing array 34 kb per patient can be analysed with a turnaround time of approximately 3 days, including analysis. Beside the high sensitivity of 96.4% to detect common missense or known mutations this methods has a high potential to detect new mutations that can not be covered by present routine techniques.