Loss of Heterozygosity of FLT3-ITD Is Common in Acute Myeloid Leukemia and May be a More Consistent Prognostic Marker Than FLT3-ITD Allele Frequency

BACKGROUND:

FLT3 alterations in Acute Myeloid Leukemia (AML) occur as either point mutations in the tyrosine kinase domain (TKD) or internal tandem duplications (ITD), both of which result in constitutive activation of FLT3; however, only FLT3-ITD has prognostic value. Per NCCN guidelines, when NPM and FLT3-ITD alterations co-occur, an allele frequency ≥ 0.5 for FLT3-ITD (FLT3-ITD^high) confers an intermediate prognosis, while allele frequencies < 0.5 (FLT3-ITD^low) confers a favorable prognosis. For allele frequency to be ≥ 0.5, loss of heterozygosity (LOH) or copy number gains at the FLT3 locus are required. In this study, we examined a large cohort of FLT3 mutated AML samples to study LOH at the FLT3 locus.

DESIGN:

During routine clinical care, 2129 AML samples were evaluated by comprehensive genomic profiling (CGP) for 406 genes via DNAseq for all classes of genomic alterations and 265 genes via RNAseq for rearrangements, using a hybrid-capture next generation sequencing assay (FoundationOne®Heme). Of these samples, 1379 met analytic specifications required for LOH analysis. LOH analysis was performed by first generating copy number models from exon and SNP log ratio and minor allele fractions (maf). LOH was then determined using the modeled copy number, maf, and tumor purity. Samples with low aneuploidy were reviewed manually.

RESULTS:

The median age of the overall cohort was 60 y (range <1y-88y) with 43.4% females and 56.6% males. Of 1379 samples, 265 (19%) had at least one functional alteration in FLT3. There were 171 (12.4%) samples with FLT3-ITD alteration(s), 115 (8.3%) samples with FLT3 TKD alteration(s), and 21 (1.5%) samples with both FLT3-ITD and TKD alterations.

LOH analysis was performed for 236 AML samples with a FLT3 alteration and 224 randomly selected wild-type FLT3 AML samples as controls. Samples with a FLT3 alteration had a significantly higher rate of LOH compared to the control samples (34/236 (14.4%) versus 2/224 (0.9%), p = 9.0x10^-9). Samples with a FLT3-ITD were under significantly greater LOH than samples with a point mutation (33/171 (19.3%) versus 4/115 (3.5%), p = 2.8x10^-5). Samples with only a point mutation rarely had LOH (1/94, 1.0%), not significantly different than FLT3 wild-type controls.

Copy number analysis revealed that FLT3-altered samples had a rate of FLT3 copy number gains not significantly different than control samples (10/236 (4.2%) versus 7/224 (3.1%)).

The allele frequency was significantly higher for FLT3-ITD alterations compared to FLT3-TKD alterations (median 0.11 versus 0.24, p < 0.001), and FLT3-ITD alterations under LOH had a higher allele frequency than those not under LOH (median 0.35 versus 0.24, p< 0.01). Only 1/115 FLT3-TKD alterations had an allele frequency > 0.5 versus 7/171 of the FLT3-ITD alterations. Five (of 7) FLT3-ITD^high samples were under LOH and the remaining 2 samples had FLT3 copy number gains.

FLT3-ITD mutated AML has co-occurring genomic alterations of clinical significance; however, with this sample size there were no statistically significant differences in the frequencies between samples with and without LOH at the FLT3 locus (NPM1: 64% vs 43%, DNMT3A: 48% vs 37%, WT1: 36% vs 25%, RUNX1: 12% vs 17%, NRAS: 3% vs 16%).

CONCLUSIONS:

In AML, the FLT3 locus has increased LOH when a FLT3-ITD is present, compared to FLT3 wild-type controls and samples with FLT3-TKD alterations. Copy number alterations in FLT3 are not different in FLT3-altered vs FLT3 wild type AML. Allele frequencies were higher for FLT3-ITDs compared to FLT3-TKDs and were highest for FLT3-ITDs under LOH.

An emerging negative prognostic indicator in AML is FLT3-ITD^high. Determination of FLT3-ITD^high status requires high tumor purity, copy number gains, and/or LOH. The requirement of high tumor purity makes FLT3-ITD^high status dependent on factors other than tumor biology, such as tumor sampling. In our dataset, 33 (19%) FLT3-ITD samples were under LOH but only 5 (4%) were FLT3-ITD^high. LOH of FLT3-ITDs has also been associated with worse prognosis in AML and further studies are warranted to determine if allele frequency or absence of wild-type FLT3 drives these prognostic correlations. FLT3-ITD LOH is more common than FLT3-ITD^high and could provide more accurate, sample independent risk stratification for patients with FLT3-ITD+ AML.