Introduction

Myeloid malignancies are clonal disorders of hematopoietic stem and progenitor cells that include myelodysplastic syndromes (MDS), myeloproliferative neoplasms (MPN) myelodysplastic/myeloproliferative (MDS/MPN) overlap neoplasms, and acute myeloid leukemia (AML). Next generation sequencing (NGS) studies have identified a number of recurrently mutated genes that have diagnostic and/or prognostic significance in these disorders. Chromosomal copy number variations (CNVs) including deletions at 5q, 7q, 12p and 17p as well as trisomy 8, are another major type of recurrent genetic alteration with clinical significance in myeloid malignancies. Detection of CNVs has traditionally required specialized testing methods such as cytogenetics/FISH and/or array-based platforms. Thus, comprehensive genetic profiling of myeloid malignancies requires multiple testing strategies at high cost. In an effort to provide more efficient genetic profiling of these disorders, we designed and tested an algorithm to evaluate for CNVs using sequence coverage data derived from a NGS-based 53-gene myeloid mutation panel with the goal of obtaining information on both mutations and CNVs from a single test.

Methods

The sample cohort included 73 MDS patients, 36 patients with MDS/MPN neoplasms, 70 MPN patients, and 91 AML patients (n=270 total cases). Genomic DNA was extracted from bone marrow or peripheral blood, and enriched for regions of interest by solution capture (SureSelect, Agilent), then sequenced on the Illumina MiSeq, HiSeq 2000 or NextSeq NGS platforms. Gene variants were identified using the software programs FreeBayes, for single nucleotide variants and small insertions/deletions, and Pindel for larger insertions/deletions. To detect CNVs in the targeted regions, the read coverage data was normalized to a Log2 ratio which was generated by comparing the normalized sample coverage to that obtained from a pool of normal controls. CNVs were detected using a circular binary segmentation algorithm. In a subset of cases (n=43) CNVs detected using NGS data were validated by comparing to the results obtained by SNP microarray (CytoScan HD Array, Affymetrix) testing, the current gold standard, and analyzed by CHAS 2.0 (Affymetrix) and Nexus 7.5 (Biodiscovery). KMT2A (MLL) partial tandem duplications detected by NGS analysis were confirmed by quantitative PCR. Comparisons of proportions were performed by Fisher's exact test.

Results

In the entire cohort of 270 cases, we detected pathogenic mutations in 208 cases (77%). ASXL1 (n=64), SRSF2 (n=40), TET2 (n=39) and DNMT3A (n=37) were among the most frequently mutated genes as has previously been shown. For targeted CNV analysis, seven cases were excluded due to inadequate normalization of the read coverages. In the validation set of 43 cases, all of the targeted CNVs detected by NGS were confirmed by SNP microarray analysis (Figure 1A). Overall, we detected targeted CNVs in 68 cases (25.8%; AML n=32, MDS n=16, MDS/MPN n=9, MPN n=11). The most frequent CNVs were 7q deletion of a region including the genes LUC7L1 and EZH2 (n=21), TP53 deletion (n=9), ETV6 deletion (n=8), gain of RAD21 (possible trisomy 8) (n=8), and 5q deletion of a region including the genes NSD1 and NPM1 (n=4). In addition, we were able to detect exon-level duplications, the so-called KMT2A partial tandem duplication (also known as MLL -PTD), in 9 cases (Figure 1B). In the 63 cases that were negative by mutation analysis (MDS n=26, AML n=17, MDS/MPN n=5, MPN n=15), targeted CNVs including 7q deletion were observed in 4 cases (6%) (MDS n=3, AML n=1). In addition, targeted CNV analysis detected TP53 deletion in 3 TP53 -non-mutated cases and in 6 TP53 -mutated cases, and TET2 deletion in 2 TET2 -non-mutated cases and in 2 TET2 mutated cases. To investigate the association among gene mutations and targeted CNVs, we found that ETV6 deletion was strongly associated with TP53 alterations (both mutation and gene deletion; p<0.001) and 7q deletion was associated with mutations in TP53, KRAS and IDH1 (p= 0.000073, 0.009, 0.026, respectively).

Conclusion

Our results demonstrated the feasibility of using the same NGS data to detect both somatic mutations and targeted CNVs with enhanced efficiency and potentially lower costs compared to classical methods.

Figure 1.

Examples of targeted CNVs detected by NGS and comparison to SNP microarray analysis.

Figure 1.

Examples of targeted CNVs detected by NGS and comparison to SNP microarray analysis.

Close modal
Disclosures

South:Affymetrix: Consultancy, Honoraria; ARUP Laboratories: Employment; Lineagen Corporation: Consultancy; Illumina: Consultancy, Honoraria.

Author notes

*

Asterisk with author names denotes non-ASH members.

Sign in via your Institution