Rapid and Reproducible Karyotyping with Nanopore Sequencing in AML Patients

Introduction: We evaluated low coverage whole genome sequencing (WGS) of acute myeloid leukemia (AML) patients using long read Oxford Nanopore Technology (ONTseq) for karyotyping and compared the results and the frequency of patients with myelodysplasia-related cytogenetic abnormalities according to previous and current AML classifications with conventional cytogenetics (CG).

Methods: Diagnostic samples from peripheral blood or bone marrow of 100 AML patients were sequenced in 3 medical centers on a GridION sequencer using SQK-LSK109 kits according to the manufacturer’s instructions, and analyzed by CG. Ten samples were sequenced independently in two different laboratories. Fifty samples were sequenced prospectively from newly diagnosed de novo AML patients. Sequencing reads were analyzed by an in-house bioinformatics pipeline based on publicly available tools such as nglmr aligner, samtools, igv_tools, R package ACE, the CyDAS webserver, and custom scripts for transforming ACE copy number models to karyotype formulas. CNVs from ONTseq data were mapped to 290 chromosome bands to allow comparison with CG. We compared the following cytogenetic categories: Complex karyotype (cKT) ≥ 3 chromosomal abnormalities; monosomal karyotype (mKT) as defined by Breems et al.; karyotype with myelodysplasia-related cytogenetic abnormalities that defines MRC-AML according WHO 2016 classification (MRC-KT), and ICC 2022 classification (MRCA-KT).

Results: ONTseq of 100 AML patients with a 4.4-fold genome coverage (range: 2.1-6.2) allowed us to resolve CNVs at a high resolution of ≥0.1 mega-basepairs (Mbp). Analysis of the ten duplicate samples established a high reproducibility of ONTseq. 110 CNVs were detected by both laboratories at identical chromosome positions and similar length (R²=0.999). Only 8 CNVs (7%) differed considerably in length between the replicates (median length difference 2 Mbp, range 0.4-16 Mbp).

Of 100 patients with a median age of 65 years (range 18-97), ≥1 chromosome band was gained or lost in 69 patients according to ONTseq analysis, while CNVs or balanced translocations were found by cytogenetics in 72 patients. Five patients had a normal karyotype by ONTseq, but not by CG, while two patients had a normal karyotype by CG but not by ONTseq. Two of these 7 discrepant patients had translocations as their sole cytogenetic abnormality [t(8;21) and t(9;11)], which cannot be detected by low coverage genome ONTseq. Patients with chromosomal abnormalities had a median of 29 (range 1-140) and 30 (range 3-136) chromosomal bands with CNVs according to ONTseq and CG, respectively. Considering all patients, gains and losses at the level of chromosomal bands correlated very well (gains: R²=0.922; losses: R²=0.923) (Figure 1).

All patients were then classified according the definitions for cKT, mKT, MRC-KT, and MRCA-KT (Table 1). Overall, patients were similarly classified by ONTseq and CG results (P<0.001). MRC and MRCA-KTs were consistently identified by ONTseq and CG in 57 patients. In 2 and 2 patients, MRC and MRCA-KTs, respectively, were identified by CG only. In 1 and 1 patient, MRC and MRCA-KT, respectively, was identified by ONTseq only. The largest discrepancies were found for mKTs due to the higher resolution of ONTseq, which more often identified subchromosomal instead of whole chromosome CNVs compared to CG, which were not considered for mKT.

Fifty of the 100 patients were sequenced prospectively at time of diagnosis. The median age of these patients was 70 years (range 61-97) and all patients had de novo AML. At least one CNV was detected by ONTseq in 24 (48%) and by CG in 26 (52%) patients. cKT, mKT, MRC-KT, and MRCA-KT were detected by ONTseq in 12, 7, 17, and 17 patients and by CG in 14, 11, 17, and 17 patients, respectively.

The median time for library preparation for prospectively analyzed patients was 3:45 hours, for ONTseq 24 hours and for bioinformatic analysis 6:30 hours. In this setting, clinical reports can be delivered on the fourth day of sample receipt.

Conclusion: Low coverage WGS with ONT provides a highly reproducible method for rapid karyotyping with high resolution. The identification of cKT and myelodysplasia-related cytogenetic abnormalities correlated very well with conventional cytogenetics, even though balanced translocations cannot be detected by our ONTseq protocol. Low coverage WGS is a promising approach for diagnostic karyotyping in AML patients.