Long-read whole genome sequencing improves clinical management in acute leukemia Free

Introduction Despite advances in molecular diagnosis of Acute Leukemia, current workflows often miss complex structural variants and pharmacogenomic markers relevant for individualized therapy. To address these limitations, rapid, long-read sequencing protocols have emerged as promising candidates for solving complex patient cases. In this analysis, we build on the foundational work of the Acute Leukemia Methylome Atlas-ALMA (Marchi et al., Nat Commun, 2025), which enabled robust epigenetic classification of acute leukemia subtypes and supported clinical outcome prediction in AML using epigenomic risk models. Here, we present a rapid and robust pipeline leveraging long-read whole genome sequencing to modify clinical management in acute leukemia patients, through comprehensive genomic and pharmacogenomic profiling.

Methods A cohort of leukemia patients from UF Health Shands Hospital, spanning pediatric and adult age groups (0.65–78 years), was evaluated using 220uL of peripheral blood (PB) and/or bone marrow (BM) specimens. A total of 33 samples from 28 patients were considered for analysis, comprising 20 BM and 13 PB. 5 samples were excluded from the analysis due to insufficient coverage. The samples failed most likely due to prolonged/compromised storage leading to DNA fragmentation. Neoplastic cells ranged from <2% to >90%, reflecting sample collection at varying time points during disease. Custom high molecular weight DNA libraries (up to 800 ng) were prepared and sequenced on a PromethION P2 Solo. Simple variants were detected by Clair3 v1.1.0, structural variants including inversions, duplications, and translocations by Sniffles2 v2.6.1, copy number variations were called with QDNAseq v1.34.0 and/or Spectre v0.2.2, and cytosine modifications by dorado v0.9.0 concomitantly without amplification or bisulfite conversion. Genotype and phenotype of pharmacogenes were called using PharmCAT v3.0.1. For each patient, an automatic clinically actionable report was generated containing ClinVar v202507 pathogenicity and CPIC/DPWG Guideline recommendations.

Results Our pipeline enabled cost-effective 5-25x long-read whole genome sequencing allowing both DNA methylation and genomic data capture to reveal clinically actionable genomic and pharmacogenomic findings within 48 hours from specimen to reporting. We present 2 representative patient cases to illustrate the clinical applicability of our integrative approach. In case 1, a 65-y.o. patient diagnosed with AML, both epigenomic and genomic data confirmed NPM1 mutation, pharmacogenomic profiling identified a poor function phenotype of the ABCG2 (BCRP) transporter, which had not been tested prior to the prescription of allopurinol at a dose of 300 mg daily, administered for 7 days on two separate occasions. In case 2, a 77-y.o. patient presenting with AML secondary to myelodysplastic syndrome, the pipeline identified a ClinVar-classified pathogenic single nucleotide variant in the IDH1 gene (c.394C>T; p.Arg132Cys), which opens eligibility for IDH1 inhibitors.

Conclusions According to DPWG guidelines, reduced ABCG2 activity is associated with decreased allopurinol efficacy, and a dose increase to 1.4 times the standard dose is recommended. The identification of this phenotype highlights the potential for personalized adjustments in clinical management during supportive care in hematologic malignancies. The patient in case 2 did not receive targeted therapy with IDH1 inhibitors, as treatment occurred prior to regulatory approval. However, study participation could have provided access to targeted therapy and potentially altered the clinical course. While current sequencing costs remain a limitation, this workflow consolidates numerous tests at once, thereby streamlining processes and substantially reducing overall complexity and associated costs. Our findings underscore the value of ~20x long-read whole genomic and epigenomic profiling for refined diagnosis, risk stratification, and supportive care management in hematology/oncology. Overall, we show the strength of a rapid long-read sequencing to provide in a single run methylation data that can assist in preliminary diagnostic and prognostic calls and DNA sequencing data for actionable genetic and pharmacogenomic calls to promote personalized treatment strategy.