Bone marrow aspiration samples were obtained from patients with leukemia. The samples were frozen in 10% dimethyl sulfoxide (DMSO) and 90% fetal calf serum in liquid nitrogen (Yu et al., 2022). SVs were detected in these samples using the following tools. Karyotyping was performed by standard G-banding methods. MLPA kits P036, P327 and P335 (MRC-Holland, Amsterdam, Netherlands) were used for B-cell leukemia samples: P036 for subtelomeric regions; P327 for the ERG gene and intrachromosomal amplification of chromosome 21; and P335 for the EBF1, IKZF1, CDKN2A, CDKN2B, PAX5, ETV6, RB1 and BTG1 genes. MLPA kits P383 (MRC-Holland, Amsterdam, Netherlands) was used for T-cell leukemia sample for STIL::TAL1, LEF1, CASP8AP2, MYB, EZH2, MLLT3, MTAP, CDKN2A/B, NUP214::ABL1, PTEN, LMO1, LMO2, NF1, SUZ12, PTPN2 and PHF6. Gene fusion events were detected by either targeted RT‒PCR or RNA-seq. Peripheral blood samples from five healthy donors were used as controls. The study received approval from the Institutional Review Board of National Taiwan University Hospital.

In this study, peripheral blood cells were used as controls, while bone marrow cells were utilized for leukemia samples. The selection of sample types was guided by the clinical relevance of these sources: peripheral blood cells (n = 5) provide a non-invasive and readily available source for controls, while bone marrow cells (n = 5) are the standard diagnostic sample for leukemia due to their higher yield of leukemic cells.

Ultrahigh molecular weight DNA was extracted from 1.5 million cells. Peripheral blood samples were used within 48 h of collection, and after red blood cell (RBC) removal (RBC lysis buffer, Qiagen), DNA was extracted with a Bionano Prep™ kit (Bionano Genomics). Frozen bone marrow samples were thawed in a 37 °C water bath and washed three times with a 10% DMSO solution, after which DNA was extracted with a Bionano Prep SP Frozen Human Blood DNA Isolation kit (Bionano Genomics). Extracted DNA was quantitated using a Qubit fluorometer.

DNA DL-green fluorophore labeling was performed using the Bionano Direct Label and Stain kit. After washing out excess fluorophores, the labeled DNA was loaded on a Saphyr Chip^® and analyzed on a Bionano Saphyr system (Bionano Genomics). Optical images of the labeled molecules were used to generate rare variant pipeline assembled genome maps with the default setting of the Bionano Solve pipeline (Bionano software v3.7). All SVs (hg38), including deletions, insertions, inversions, and translocations, were annotated with the Variant Annotation Pipeline.

WGS was conducted on an Illumina NovaSeq 6000 system with an average coverage depth of 30X. The raw sequencing reads were aligned to the hg38 reference genome using the BWA-GATK-ANNOVAR pipeline. SVs identified by the Manta program were searched for in the gaps left by OGM alignments.

The statistical analyses were performed using SPSS software (version 25.0 and 22.0; IBM Corp., Armonk, NY, United States). Variables related to leukemia group and the control group were analyzed using the Mann‒Whitney U test for comparison. A p value <0.05 was considered to indicate statistical significance.

OGM analysis of the five leukemia samples achieved an effective coverage of >300x in all samples, with an average label density of 15.07 per 100 kb (SD 0.93) and a mapping rate of 87.8% (SD 4.49). After filtering, an average of 1,044 SVs were identified in each sample, including 477 insertions, 457 deletions, 32 inversions, and 73 duplications (Supplementary Table S1). In comparison, an average of 650 SVs were identified in the control samples, including 315 insertions (p = 0.016), 284 deletions (p = 0.028), 32 inversions (p = 0.028), and 17 duplications (Supplementary Table S1). Chromosomal translocation (intertranslocation) was observed only in leukemia samples.

OGM detected all previously known SVs except for the IGH::DUX4 fusion detected by RNA-seq in Case 3. OGM also detected additional SVs in these samples (Table 1).

Case 1

Conventional tests yielded a karyotype of 46,XX,t(5;8)(q31;q24),?der(12)t(11;12)(q13;p11),idic(21)(p11)[8]/46,XX[12]. The ETV6::RUNX1 fusion (t(12;21)(p13;q22) (exon 5-exon 3)) was detected by RT‒PCR, and the 21q11.2-22.12 duplication, ETV6 exon 8 deletion, and ETV6 exons 1–5 duplication were detected by MLPA. OGM analysis revealed a more complicated karyotype, including sequential translocations between chromosomes 5, 8, 12, and 21 at 5q23, 8q22.3, 12p12.1, and 21q22 (Figures 1, 2). Additionally, a deletion in chromosome 12 (p12.1 to p13.2) was detected only by OGM. The changes in chromosome 21 are more complicated than those observed by conventional karyotyping; for example, 12p13.2 was first translocated to 21q22, and 21q22 was subsequently translocated to 5q23; second, the derived 21q was duplicated, which led to the formation of a dicentric chromosome 21. OGM also detected a translocation between the termini of chromosomes 1 and 2. All gaps at breakpoints mapped by OGM were closed by WGS (Supplementary Table S2), including the BCAT1::BAALC fusion between chromosomes 8 and 12.

Case 2

Conventional tests yielded a karyotype of 46,XX in 20 cells, BCR::ABL1 fusion (t(9;22)(q34;q11) (exon 1-exon 2)) by RT‒PCR and deletions of IKZF1 exons 2–7 (7p12.2), CDKN2A/2B (9p21.3), and PAX5 exons 7–10 (9p13.2) by MPLA. OGM revealed a translocation between chromosomes 9 and 22, which explained the BCR::ABL1 fusion. OGM revealed a large deletion of chromosome 9p (p13.3 to p21.3), which encompassed 16,321,476 base pairs and contained 245 genes, including CDKN2A/2B and PAX5. OGM further revealed a smaller deletion of 88,550 base pairs on chromosome 7 involving IKZF1. The large deletion (>16 Mb) identified by OGM may have been missed by karyotyping due to the mosaic nature of the cancer cells (with a deletion heterozygosity of approximately 20%) or their reduced proliferative capacity in standard culture conditions. WGS defined the breakpoint for the Philadelphia chromosome as chr9:130813896 fused with chr22:23227330 and chr22:23227334 fused with chr9:130813908.

Case 3

The only previously known abnormality in this patient was the IGH::DUX4 fusion with the highly expressive of DUX4 gene detected by RNA-seq (Yu et al., 2022). OGM revealed a terminal translocation between chromosomes 6p25.2 and 14q32.3 and an IGH::DUSP22 fusion (Figures 3A,B). This gene fusion has been associated with chronic myeloid leukemia and lymphoma. (Savage and Slack, 2023; Pedersen et al., 2017) We could not confirm the fusion of IGH::DUX4. There were few OGM probes in this region, and WGS could not map this region because of the presence of multiple DUX4 pseudogenes (Figures 3C,D).

Case 4

Conventional tests yielded a karyotype of 46,XX,inv(16)(p13q22), and the CBFB::MYH11 fusion (chr16:15721182-67089716) was detected by RT-PCR. OGM/WGS identified both the breakpoint of the MYH11::CBFB fusion and 676 genes in the inversion region. Since Case 4 is AML, and prognostic factors for AML are typically SNVs, MLPA was not performed.

Case 5

Conventional karyotyping results were normal, an STIL::TAL1 fusion was detected by RT‒PCR, and CDKN2A/2B (9p21.3) and STIL (1p33) deletions were detected by MPLA. OGM determined that the STIL::TAL1 fusion on chromosome 1 was caused by an 81,839 bp deletion, and another 115,539 bp deletion was detected on chromosome 9, which involved CDKN2A/2B.

In our study, we found that the leukemia samples had an average of 477 insertions, 457 deletions, and 32 inversions, which were significantly greater than those observed in the normal blood samples (p = 0.016, 0.028, and 0.028, respectively). Fresh peripheral blood was used as controls, while frozen bone marrow was used for patients. Although freeze-thaw may cause random DNA breakage, the accuracy of OGM was not affected because DNA fragment lengths were sufficiently long in all samples.

The current study demonstrated the power of OGM as a single test for detecting SVs in leukemia patients. Cytogenetic studies have been shown to be crucial for identifying genes implicated in the development of human leukemia (Nowell, 1993). However, only a portion of gene fusion events can be detected by karyotyping alone. FISH can detect a panel of chromosomal aberrations but requires time-consuming techniques (Neveling et al., 2021). Microarrays are an efficient alternative to karyotyping in leukemia diagnosis, which is mainly associated with SVs (Neveling et al., 2021). Therefore, the most common practice is to design a panel for RT‒PCR or MLPA to detect known SVs involved in leukemia (Smeets, 2004; Vissers et al., 2005). In contrast, OGM efficiently detected SVs in a straightforward way. It would be useful since high hyperdiploid is the most common subtype of leukemia. The cost of OGM is probably similar to the combined cost of conventional tests but less than that of long-read sequencing.

In Case 1, OGM detected the BCAT1::BAALC fusion. Although the BCAT1::BAALC fusion has not been previously reported, both BCAT1 and BAALC are associated with chronic myeloid leukemia (Hattori et al., 2017). BCAT1 catalyzes the initial step of branched-chain amino acid catabolism and is known to promote cancer proliferation and invasion through the activation of either the phosphatidylinositol 3-kinase/protein kinase B/mammalian target of rapamycin pathway or Wnt/β-catenin signal transduction (Nong et al., 2022). BAALC expression has also been associated with acute lymphoblastic leukemia and acute myeloid leukemia (Tanner et al., 2001). In Case 3, OGM detected the IGH::DUSP22 fusion. DUSP22 is a tumor suppressor gene, and rearrangements in this gene have been associated with favorable outcomes in patients with kinase-negative anaplastic large cell lymphoma (Pedersen et al., 2017). DUSP22 is located at the end of chromosome 6p, a location difficult to identify by karyotyping. The translocation of the IGH proto-oncogene is a common driver event in leukemia (Tian et al., 2019). The significance of these findings is further underscored by recent observations linking monoallelic 6p25.3 rearrangements with DUSP22 to lymphoma and leukemia (Savage and Slack, 2023; Arruga et al., 2017). The multiplex reciprocal translocation in Case 1 is intriguing. This variant likely involved sequential translocations of chromosomes 5, 8, 12, and 21. The four chromosomes might be organized together in the nucleus, so the four translocations could occur simultaneously. The use of OGM to detect new fusion genes can provide invaluable insights for cancer research.

OGM detected all SVs known from conventional tools in Case 2, Case 4, and Case 5. OGM also revealed other SVs currently lack established clinical significance, but these information could be valuable for future researches. The clinical implications of these results obtained by OGM are multifaceted. First, OGM offers higher resolution for SVs than by conventional tools like karyotyping or FISH. This leads to improved diagnostic accuracy, which is crucial for personalized medicine. Second, OGM enables the identification of novel or rare SVs, enabling understanding of disease mechanisms or finding new biomarkers useful for classification or prognosis. Third, OGM generates comprehensive SV information that can be generated by multiple conventional tests. Therefore, OGM simplifies diagnostic workflows, reduces turnaround times, and may lower overall costs, making it a powerful tool for advancing precision medicine.

OGM can detect chromosomal translocations, deletions, duplications, and inversions throughout the genome. However, because OGM depends on the location of fluorescence markers within the genome, the exact sequences at the breakpoints have not been resolved. For novel gene fusion events or fusion events in which the reading frames of the downstream gene are important, it may be important to sequence the breakpoint. This can be accomplished by designing breakpoint PCR and Sanger sequencing methods. In the present study, we employed short-read WGS to close the gap in the breakpoint left by OGM. Short-read WGS is thought to have low sensitivity for SVs, and most tools usually generate tens of thousands of SVs from a single analysis; these SVs are almost impossible to interpret (Nakagawa and Fujita, 2018). However, in the present study, when we had already located the breakpoint within a few kb of sequences by OGM, it was quite easy to identify the breakpoint read by WGS and precisely define the breakpoint, including the read frames. Because WGS is inexpensive and quick, it is not a burden to add WGS to OGM, and the entire process can be completed in days. Therefore, short-read WGS effectively close the gap at the breakpoints of SVs detected by OGM. However, short-read WGS itself, or conventional WGS, does not unambiguously detect SVs. A key advantage of OGM/WGS is their ability to bypass the need for cell culture. Karyotyping and FISH typically require 14 days to 1 month for cell culture. While MLPA and RT-PCR are fast, they only target specific regions (Supplementary Table S3). In recent years, artificial intelligence (AI) has aided both conventional cytogenetics and new technologies like OGM to improve cancer cytogenetic analysis (Alain, 2024).

In Case 3, the IGH::DUX4 fusion previously detected by RNA-seq could not be confirmed by either OGM or WGS. DUX4 is located at the terminal end of chromosome 4 and contains 11 to more than 100 D4Z4 repeats (Dib et al., 2019). Unfortunately, OGM coverage in this particular region was low, and this region was also refractory to short-read sequencing alignment. Recently, long-read sequencing has been shown to detect D4Z4 repeat contraction in patients with facioscapulohumeral muscular dystrophy (FSHD) (Pearson, 2010; Yeetong et al., 2023). Therefore, long-read sequencing may be needed to confirm the IGH::DUX4 fusion.