Comprehensive Genetic Diagnostics of Acute Myeloid Leukemia By Next Generation Sequencing

Background

Acute Myeloid Leukemia (AML) is the most common acute leukemia in adults with a poor overall prognosis. Although the disease has been extensively characterized on the molecular level, this knowledge is translating only slowly into the clinic, particularly with regard to novel therapeutic concepts. Presumably, this striking imbalance substantially is due to the long time required to complete genetic analyses so that results are not available when treatment has to be initiated. Specifically, cytogenetic examinations to determine the karyotype of the malignant blasts, which has been the most important parameter for risk stratification for more than thirty years, take up to two weeks. Next generation sequencing (NGS) technology essentially catalyzed efforts to dissect the genomic landscape of AML, leading to the identification of a large variety of AML driver genes and distinct molecular risk groups. However, these emerging molecular classes of AML do not cover all patients, implying that karyotyping is not dispensable for AML diagnostics at this point. Here we present an integrated approach to AML diagnostics that incorporates these complementary genetic examinations - focused mutational screening of AML-related genes and karyotyping - in one NGS assay.

Methods

We combined targeted resequencing of DNA and RNA using commercially available panels (TruSigth Myeloid, Illumina and FusionPlex Heme, ArcherDx) to detect AML-associated short sequence variants and gene fusions with low coverage whole genome sequencing for copy number variation analysis. Sequencing was performed on an Illumina MiSeq instrument with a read length of 2x150 bp and a coverage of 3.75 M reads for the TruSight Myeloid panel, 2.25 M reads for the FusionPlex panel and 1.5 M reads for the whole genome library. Variants and fusions were called using the manufacturers' analysis software and a previously published algorithm to identify ITDs (ITD-seek, Au et al., 2016). CNV analysis was performed by comparing read distribution in an AML whole genome library to in silico randomly sampled reads from the reference genome using an in house-developed algorithm.

Results

Initial testing of our approach on leukemia cell lines and peripheral blood leukocytes from healthy donors revealed sensitivities of 2% and 1-25% for the detection of DNA variants and fusions, respectively. Applying stringent filter criteria, we recovered 75% of verified COSMIC variants and 100% of known fusions in undiluted AML samples without false positives. Chromosomal gains and losses were detected with high confidence with a sensitivity of 10%. We were able to reliably distinguish between normal and complex karyotypes, although NGS-karyotyping based on known fusions and CNV-analysis missed some details of highly aberrant karyotypes such as derivative chromosomes and chromosomal translocations that did not involve genes included in the FusionPlex panel. Our preliminary experience on our method in a diagnostic setting confirms high correlation with reference laboratory results and no relevant differences with regard to treatment decisions. Moreover, we find that NGS considerably accelerates genetic diagnostics of AML as the entire workflow from sample to report including three parallel library preparations, sequencing and data analysis can be completed within 5 days. Operational costs amount approximately 1,700 USD (1,500 EUR) per sample with the low throughput equipment used in this work, which is in the range of expenses for currently established AML diagnostics.

Conclusions

NGS allows for comprehensive translocation and mutation screening, however, some technical and bioinformatics optimization is required to achieve consistently high sensitivity and specificity for all target genes. CNV analysis of low coverage whole genome sequencing data adds valuable information on numerical chromosomal aberrations, thus allowing construction of a virtual karyotype to substitute for difficult and time-consuming cytogenetics. In summary, we present a reliable, fast and cost-effective strategy to combine molecular and cytogenetics for AML diagnostics in a single NGS run in order to pave the way for a more differentiated clinical management of AML patients in the near future.