Differentiating between Hyperdiploidy and Pseudo-Hyperdiploidy in B-Lymphoblastic Leukemia Utilizing Low-Coverage Mate-Pair Sequencing

Purpose:

B-cell acute lymphoblastic leukemia/lymphoma (B-ALL/LBL) represents the most common childhood malignancy. Classification of recurrent genetic abnormalities in B-ALL/LBL is essential for risk stratification and patient management. Observed in approximately 5% of B-ALL/LBL cases, hypodiploidy is considered a poor prognostic finding and can be further characterized as near-haploidy (24-31 chromosomes), low hypodiploidy (32-39 chromosomes) and high hypodiploidy (40-43 chromosomes). Importantly, masked hypodiploid clones that have undergone endoreduplication can be mischaracterized as hyperdiploidy, a favorable prognostic finding in B-ALL/LBL. These "doubled" hypodiploid clones are termed pseudo-hyperdiploid and their accurate detection is critical.

Mate-pair sequencing (MPseq) is a next-generation sequencing (NGS) technology designed to sequence larger DNA fragments (2000-5000bp) than paired-end sequencing. This allows for the detection of large genetic rearrangements (e.g. translocations, inversions, etc.) and copy number variants (CNVs) at a high resolution with low base coverage. It has demonstrated particular clinical utility for genetic characterization of hematologic malignancies. To expand the role of MPseq in the evaluation of hematologic neoplasms, we aimed to improve the analysis of MPseq results to allow for the detection of copy-neutral loss of heterozygosity (cnLOH) that would enable the differentiation between hyperdiploid and pseudo-hyperdiploid clones.

Methods:

A new algorithm was applied that statistically analyzes common SNP locations in MPseq data in order to detect both cnLOH regions, and provides global heterozygosity levels at a resolution that allows for the detection of hypodiploidy. This method accounts for the low base-coverage of MPseq data (<10x) in order to provide the highest possible accuracy and precision without increasing cost by requiring changes to the input or sequencing requirements. The algorithm was applied to 17 B-ALL samples classified by conventional chromosome analysis, FISH, and/or chromosomal microarray as either pseudo-hyperdiploidy (n=8) or hyperdiploidy (n=9). The cases were sequenced using MPseq with 101bp read lengths across a range of base coverages (1-10x; 30-135 million fragments) and tumor percentages (50-100%).

Results:

We demonstrate the method's ability to differentiate between hyperdiploidy and pseudo-hyperdiploidy in all cases without any changes to the MPseq input or sequencing requirements. We detected an average of 11.5 whole-chromosome cnLOH (range 8-20) in the pseudo-hyperdiploidy cases, compared to an average of 0.67 whole-chromosome cnLOH (range 0-2) in the hyperdiploidy cases. Additionally, the method allows for the detection of cnLOH regions across the genome. For example, we observed a recurrent cnLOH of chromosome 9 in the hyperdiploid subtype (3/9 hyperdiploid cases), which is a recurrent abnormality in the disease. The method provides an easy to read visual output that compares favorably to the output of chromosomal microarray, increasing its utility for clinical application.

Conclusions:

This advancement in the analysis of low-coverage sequencing data makes MPseq available for the genetic characterization of B-ALL/LBL, where hypodiploidy is a primary cytogenetic abnormality and is sometimes missed by conventional cytogenetic techniques. Additionally, the algorithm allows for the detection of cnLOH regions across the genome, which is important across many disease types for diagnosis, prognosis, and the detection of potential therapeutic targets. The addition of the algorithm to the MPseq analysis pipeline makes it a more complete genetic characterization tool and brings it closer to its promise as a potential single assay replacement for conventional cytogenetic techniques.