Session: 803. Emerging Tools, Techniques, and Artificial Intelligence in Hematology: Poster I
Hematology Disease Topics & Pathways:
Research, Lymphoid Leukemias, ALL, Acute Myeloid Malignancies, AML, Translational Research, APL, Bioinformatics, Diseases, Lymphoid Malignancies, Myeloid Malignancies, Technology and Procedures, Omics technologies
Background: Long-read sequencing theoretically has a high capability for detecting structural abnormalities, making it an optimal choice for the genomic diagnosis of leukemia. Especially, the real-time nature of nanopore sequencing allows for an in-silico targeted sequencing method, targeted adaptive sampling long-read sequencing (TAS-LRS). However, the analysis pipeline is still under development, and its accuracy and characteristics are currently under investigation Leveraging the comprehensiveness and rapidity, we applied TAS-LRS to genome profiling for pediatric leukemia and evaluated its utility.
Methods: We performed TAS-LRS for 28 tumor/normal-paired samples of pediatric leukemia (10 acute myeloid leukemia [AML], 13 B-cell acute lymphoblastic leukemia [ALL], and 5 T-ALL) using a GridION sequencer and R10 flow cells. Target regions were composed of 466 genes associated with hematologic malignancies. After 3-day sequencing, single-nucleotide variants (SNVs), structural variations (SVs), and copy number variations (CNVs) were determined. In 21 samples, the results were compared with those by short-read whole genome sequencing (WGS).
Results: The process from library preparation to reporting the final results took approximately 72 hours. The median coverage and N50 was 21.0× and 11,191 bps, respectively, in the on-target regions in tumor samples. Genomic subtypes were determined in 24 (85.7%) patients by TAS-LRS. These included 12 patients whose genomic alterations had not been identified by clinical tests and all of the abnormalities were SVs, such as DUX4 rearrangement and cryptic KMT2A rearrangement. In the other four patients whose genomic alterations could not be identified by TAS-LRS, putative driver alterations that were not involved with genomic subtypes were detected in two, and germline RUNX1 deletion was detected in one. No driver alterations were detected only in one patient with AML. Both chromosomal-level and focal CNVs, such as high hyperdiploid and CDKN2A deletion, could be detected efficiently using genome-wide low-coverage reads in the off-target regions. In addition to the identification of genomic subtypes as mentioned above, the technical nature of this method could be used for other applications, including NUDT15 diplotyping and detecting accurate breakpoints of fusion genes for fusion-based minimal residual disease assay. TAS-LRS had an inferior capability for detecting SNVs and small indels compared to short-read WGS, with a sensitivity of 60.9% for the SNVs and 17.6% for the small indels among those identified short-read WGS. For example, in the patient with AML whose driver alteration could not identified by TAS-LRS, a 4-bp insertion of NPM1 was detected by short-read WGS. TAS-LRS also had poor efficiency for detecting variants with low variant allele frequency due to the low coverage. Conversely, TAS-LRS could identify 85.3% of the SVs detected by short-read WGS. Some SVs were detected only by TAS-LRS. Most of them were rearrangements in immunoglobulin/T-cell receptor genes which were overlooked by short-read WGS due to low mapping quality.
Conclusions: These findings illustrated the viability of TAS-LRS as a rapid and comprehensive genome-profiling method for pediatric leukemia, particularly advantageous for identifying SVs and CNVs. Our method would contribute to precise diagnosis and classification, pharmacogenomic genotyping, and enhancement of minimal residual disease assay in pediatric leukemia.
Disclosures: No relevant conflicts of interest to declare.
See more of: Oral and Poster Abstracts