Comprehensive genetic diagnosis of acute myeloid leukemia by next-generation sequencing

Abstract

Differential induction therapy of all subtypes of acute myeloid leukemia other than acute promyelocytic leukemia is impeded by the long time required to complete complex and diverse cytogenetic and molecular genetic analyses for risk stratification or targeted treatment decisions. Here, we describe a reliable, rapid and sensitive diagnostic approach that combines karyotyping and mutational screening in a single, integrated, next-generation sequencing assay. Numerical karyotyping was performed by low coverage whole genome sequencing followed by copy number variation analysis using a novel algorithm based on in silico-generated reference karyotypes. Translocations and DNA variants were examined by targeted resequencing of fusion transcripts and mutational hotspot regions using commercially available kits and analysis pipelines. For the identification of FLT3 internal tandem duplications and KMT2A partial tandem duplications, we adapted previously described tools. In a validation cohort including 22 primary patients' samples, 9/9 numerically normal karyotypes were classified correctly and 30/31 (97%) copy number variations reported by classical cytogenetics and fluorescence in situ hybridization analysis were uncovered by our next-generation sequencing karyotyping approach. Predesigned fusion and mutation panels were validated exemplarily on leukemia cell lines and a subset of patients' samples and identified all expected genomic alterations. Finally, blinded analysis of eight additional patients' samples using our comprehensive assay accurately reproduced reference results. Therefore, calculated karyotyping by low coverage whole genome sequencing enables fast and reliable detection of numerical chromosomal changes and, in combination with panel-based fusion-and mutation screening, will greatly facilitate implementation of subtype-specific induction therapies in acute myeloid leukemia.

Figure 1.

Comprehensive genetic diagnosis of acute myeloid leukemia by next-generation sequencing. (A) Outline of the workflow: each sample is subjected to preparation of three sequencing libraries. Libraries are indexed separately for sequencing on the same flowcell. Data are analyzed using five distinct algorithms for the detection of CNV, fusions, and DNA variants. The whole workflow can be completed within 5 days if performed by one person; times to perform individual steps of the composite assay are indicated on the right. (B) Outline of the CAI[N] algorithm for CNV analysis. Reads are mapped to 1 Mb fixed genomic windows and read distributions are compared to the average of more than 2,500 normal karyotypes (N_female=2,819, N_male=2,605) generated by random sampling of 150-250 bp reads from the reference genome. A region is called amplified or deleted if the observed read number in a window differs significantly (P<0.003) from the average of in silico-generated karyotypes. (C) Flow cell occupancy by three sequencing libraries. Two samples can be analyzed in parallel in one sequencing run on a standard MiSeq v2 flowcell when libraries are sequenced with the read numbers indicated in (A).

Figure 2.

Calculated chromosome banding and in silico-generated reference karyotypes. (A) Calculated chromosome banding by CAI[N] analysis of lc-WGS data. Read distributions in genomic windows along chromosome 9 of an AML patient (AML-2, Table 1) are indicated (left: cytogenetic bands, right: 1 Mb windows. (B) Frequencies of uniquely mapped reads on whole chromosomes for in silico-generated normal karyotypes. RF: random female (N=2,819), RM: random male (N=2,605). Error bars represent the standard deviation (below visibility; <0.01%). Note that in (A) the centromere of a chromosome is not covered and in (B) the Y chromosome appears smaller than its actual size because of repetitive DNA sequences, which prevent unique alignment of sequencing reads. (C) Scalability of the CAI[N] algorithm: Four whole genome libraries from two healthy female donors were sequenced with different read numbers in multiplexed sequencing runs (right panel). Healthy F1.1-4: four runs of the same library.

Figure 3.

Detection of whole chromosome gains and losses by copy number variation karyotyping. Whole genome libraries from (A) an individual with Down syndrome (T21) and (B) the BEN-MEN-1 cell line were sequenced with low coverage and analyzed by CAI[N]. RF: random female (N=2,819), RM: random male (N=2,605). Error bars represent the standard deviation (below visibility).

Figure 4.

Detection of partial chromosome losses and gains by copy number variation karyotyping. Whole genome libraries from three AML patients’ samples were sequenced with low coverage and analyzed by CAI[N]. (A) Region plots for chromosome 5. (B) Region plot of chromosome 1 for patient AML-1. Read numbers in 1 Mb windows were normalized to 1 ×10⁶ total reads. RF: random female (n=2,819), RM: random male (n=2,605). See also Online Supplementary Figures S1-S3.

Figure 5.

Sensitivity of copy number variation karyotyping. Genomic DNA from the BEN-MEN-1 cell line (monosomy 22) was diluted in healthy donor DNA (Healthy F1, Figure 2) in different ratios and subjected to lc-WGS and CAI[N] analysis. (A) Region plots for chromosome 22. The range ±3 standard deviations around the mean is indicated in pale red. (B) CNV decision plots. Read numbers in 1 Mb windows were normalized to 1×10⁶ total reads. RF: random female (N=2,819). Color coding in (B) as in (A).