Detailed molecular characterisation of acute myeloid leukaemia with a normal karyotype using targeted DNA capture

Abstract

Advances in sequencing technologies are giving unprecedented insights into the spectrum of somatic mutations underlying acute myeloid leukaemia with a normal karyotype (AML-NK). It is clear that the prognosis of individual patients is strongly influenced by the combination of mutations in their leukaemia and that many leukaemias are composed of multiple subclones, with differential susceptibilities to treatment. Here, we describe a method, employing targeted capture coupled with next-generation sequencing and tailored bioinformatic analysis, for the simultaneous study of 24 genes recurrently mutated in AML-NK. Mutational analysis was performed using open source software and an in-house script (Mutation Identification and Analysis Software), which identified dominant clone mutations with 100% specificity. In each of seven cases of AML-NK studied, we identified and verified mutations in 2-4 genes in the main leukaemic clone. Additionally, high sequencing depth enabled us to identify putative subclonal mutations and detect leukaemia-specific mutations in DNA from remission marrow. Finally, we used normalised read depths to detect copy number changes and identified and subsequently verified a tandem duplication of exons 2-9 of MLL and at least one deletion involving PTEN. This methodology reliably detects sequence and copy number mutations, and can thus greatly facilitate the classification, clinical research, diagnosis and management of AML-NK.

Figure 1

Workflow diagram for data analysis and mutation calling. After initial parsing of sequencing data through a series of open source software tools, mutation calling is performed by our in-house Perl script (MIDAS). Mutational consequences are then determined by Variant Effect Predictor, Ensembl. For the purposes of comparing MIDAS with other callers, SomaticSniper and VarScan were used instead.

Figure 2

Distribution of the depth of sequencing coverage of the target genes. Representative data from sample P1 showing the fraction of bases covered at incremental depth windows (blue bars and left hand y axis) and the cumulative fraction of bases covered at or above the specified coverage (orange line and right hand y axis). This shows that ∼88% of bases were covered at by at least 1000x sequencing reads.

Figure 3

Identification of MLL partial tandem duplication (PTD) using sequencing read depth. Normalised per exon sequencing read depths for the first 14 exons of MLL show increased depth for exons 2–9 from sample P6 (a). This suggested the presence of an exon 2 to exon 9 PTD with a breakpoint in intron 9 (b). PCR amplification across the putative breakpoint using an exon 9 forward (MLL_9F) and an exon 2 reverse (MLL_2R) primer confirms the presence of the PTD in this AML sample (c).