Haloplex target DNA enrichment allows investigation of copy number and mutational status of key genes in acute myeloid leukaemia with normal karyotype

Background: Acute myeloid leukaemia (AML) is a deadly haematological malignancy for which individualized treatment based on prognostic stratifica- tion is essential to maximize chances of survival. Karyotypic abnormalities car- ry prognostic significance, but are absent in about 50% of patients (NK-AML), and in this subgroup analysis of gene mutations can be used to stratify patients. Recent advances in AML genomics have defined the set of genes which are recurrently mutated in NK-AML, whilst the prognostic impact of many of these has been determined. Therefore, clinical-grade sequencing platforms will be increasingly useful in clinical practice in the next few years, and this highlights the need for reliable methods targeted gene re-sequencing. Aims: HaloPlex is a novel, rapid approach for targeted DNA enrichment that requires low amounts of input DNA. We evaluated its performance in detect- ing abnormalities in coding sequence and copy number of genes recurrently mutated in NK-AML, with special focus on reproducibility and on the quantita- tive value of data. Methods: Genomic DNA from 43 NK-AML samples from 40 patients was sub- jected to Haloplex target enrichment for 24 genes. Target-enriched DNA was sequenced on HiSeq2000, and reads were aligned using BWA. Substitutions and indels were called using standard algorithms and mutations called as pre- viously described (Papaemmanuil et al., Blood 2013). Results: We sequenced 32.26 gigabases (Gb). A mean of 66.13% mapped on- target (62.94%>74.15%). The mean coverage of the target region was 3674.69x and 91.16% of bases were covered at >30X. Read depth showed sig- nificant variability, and coverage across consecutive bases could vary by sev- eral folds. This variability was expected as HaloPlex target-enrichment employs digestion of genomic DNA and specific capture of variable-length fragments with subsequent PCR amplification. Nevertheless, we found that this variabil- ity in coverage was predictable and highly consistent across samples. Conse- quently, using normalized data we were able to identify an interstitial deletion of BCOR and three MLL partial tandem duplications. We also report an ampli- fication of KRAS, an event with driver potential in solid cancers but not previ- ously described in AML. The quantitative nature if the data was retained when looking at point mutations, as demonstrated by a narrow range of allelic fre- quency of heterozygous SNPs. We identified likely oncogenic mutations in 38/40 samples with a median of 3 (1-5) per sample. NPM1 mutations were the most frequent (69%), followed by FLT3 (58%), DNMT3A (35%) and TET2 (32%). As described, NPM1 mutations co-occurred with FLT3 and DNMT3A. Both mono- and bi-allelic FLT3-ITDs were reliably identified. The recurrence rates were consistent with previously published data. We estimated allelic fre- quency of each mutation and reconstructed the phylogeny of mutations in each. In 2 cases with serial samples we studied the dynamics of tumour evolution and described variants lost and gained at relapse, implying that subclonal evolu- tion can be inferred using Haloplex targeted data. Summary and Conclusions: Haloplex target enrichment followed by mas- sively parallel sequencing is a simple, rapid and robust method for high through- put screening of gene alterations in NK-AML. It may help prognostic stratifica- tion, treatment decisions and minimal residual disease assessment in clinical practice. It can reliably call copy number alterations, substitutions and indels. However, the targeting design.