Loss of Heterozygosity Identified in Acute Myeloid Leukemia with Normal Karyotype Using SNP Microarrays.

Loss of heterozygosity (LOH) is detectable in many forms of cancer including leukemia and contributes to tumorigenesis through the deletion of tumor suppressor genes. It derives from the loss of one of the two alleles at a given locus caused by deletion or uniparental disomy (UPD). In this study we describe the genome wide analysis of LOH in purified leukemic blasts from acute myeloid leukemia (AML) samples with normal karyotype using a novel technique based on single nucleotide polymorphisms (SNP). In total we selected 50 peripheral blood samples of de novo AML patients with normal karyotype at the time of diagnosis. Patients were treated according to the AML-96 multi-center protocol of the DSIL. Pure leukemic cells and tumor free material (T cells) from each patient were obtained using FACS-Vantage cell-sorting (BD Sciences, Germany). DNA was isolated from sorted cells. Genome wide SNP analysis was carried out according to the standard GeneChip Mapping Assay protocol (Affymetrix, USA) with pre-amplified DNA (Repli-g^™ Kit; Molecular Staging Inc. USA) using the Human Mapping 10K Arrays XbaI 131 (Affymetrix). Individual regions of potential LOH identified by the Affymetrix® GeneChip® Chromosome Copy Number Analysis tool were confirmed by microsatellite analysis of short tandem repeat (STR) markers using the matched non-manipulated original DNA samples. Genome wide analysis of SNP in pre amplified DNA of FACS sorted cells from AML samples with normal karyotype detected long stretches of hemizygosity, indicative of LOH in 8/49 evaluable patients (16%). In 6 of these cases STR-analysis of T cells representing the corresponding tumor free material confirmed the regions of partial UPD. UPD affected four different chromosomes (chromosome 2p and 11q, in each case twice; chromosome 8q, and 13q) and covered between 11.5 and 88 Mb. To our surprise in the healthy material of the remaining two cases no heterozygote loci were identified at the affected chromosomal regions (chromosome 3 14.5 Mb; chromosome 20 29.3 Mb) and consequently identified as unusual long stretches of homozygosity present in both the malignant and the healthy cells. These cases might reflect genotypes with high susceptibility to malignant mutations. No differences were observed for any clinical factors, including age, WBC-counts, sex and FAB-subtype. Also, several of the mutations frequently identified in patients with normal karyotype (FLT3-ITD, MLL-PTD, NPM1) had a comparable prevalence in patients with and without UPD. Interestingly, although 5/6 patients with UPD achieved complete remission after induction chemotherapy, 4/5 (80%) relapsed within the first 6 months. In contrast the rate of relapse in patients without UPD was only 54% (15/28). The only patient positive for UPD and alive in remission received an allogeneic stem cell transplantation. In conclusion, the combination of whole genome amplification method and SNP array technology allows the identification and mapping of LOH in AML patients with normal karyotype. Our data also point to the necessity to analyze tumor free material to confirm the somatic origin of the alteration. Although small numbers of patients were investigated, our data might indicate that patients with UPD have a high rate of treatment failure. This should be further investigated in larger cohorts of patients.