Feasibility and utility of unmatched whole-genome sequencing (uWGS)  for pediatric acute leukemias: An analysis of leukemia precision-based therapy (LEAP) study samples

Abstract

Abstract Introduction For children with acute leukemias, rapid and accurate assessment of genomic biomarkers is essential for optimal diagnostics and therapy selection. Current approaches rely on multimodal platforms including karyotyping, fluorescent in situ hybridization (FISH) and targeted DNA/RNA sequencing (TS). We, along with other groups, have previously demonstrated the feasibility of accurately identifying clinically relevant alterations across variant classes via unmatched whole genome sequencing (uWGS) of pediatric and adult leukemia patients (Gutierrez-Abril et al., 2022; Duncavage et al., 2021). To build on this work and assess uWGS feasibility and clinical utility for pediatric leukemias, we performed uWGS on a cohort of 68 samples collected as part of the Leukemia Precision-based Therapy (LEAP) study (NCT02670525; Pikman et al., 2021). We report concordance with standard cytogenetics and TS, as well as incremental biologically relevant findings for a subset who did not have clear driver genomic alterations identified through the LEAP study. Methods We selected a cohort of pediatric patients with B- or T-cell acute lymphoblastic leukemia (B-ALL, n=36; T-ALL, n=2), acute myeloid leukemia (AML, n=26), mixed phenotype acute leukemia (MPAL, n=3), or blastic plasmacytoid dendritic cell neoplasm (BPDCN, n=1), whose bone marrow or blood specimens had either unknown or incomplete driver alteration profiles (n=43), or previously defined drivers (n=25). Twenty patients had newly diagnosed leukemia and 48 had relapsed/refractory leukemia. uWGS was performed on samples with a median coverage of 110x (range 98-139), and TS with 1349x (range 346-2072). The median purity level was 0.78 (range 0.13-1) in the cases with identified copy number variation (CNV). New uWGS findings were categorized as oncogenic, prognostic, diagnostic, and/or therapeutically relevant based on genome databases (eg, OncoKB, COSMIC) and review of published literature. Clinical karyotype, FISH, and/or previously-obtained DNA-based somatic TS data were available for all 68 samples. RNA-based fusion-panel data was also available for 40 samples (59%). Concordance was evaluated for alterations with a cancer cell fraction (CCF) for CNVs or variant allele frequency (VAF) for point mutations of at least 10%. Results uWGS results were concordant with 95% (95/100) of karyotyping, 100% (21/21) of FISH and 96% (171/178) of TS. There were 5 alterations identified by karyotype from 3 different samples not detected by uWGS: 3 derivative chromosomes (der(X)t(X;1)(q22-24;q21); der(21)t(7;21)(p12;p11.2); der(17)t(9;17)(p13;-11.2)), one inversion (inv(2)(p23q31)) and one deletion (del(3)(q22q26.2)). For mutation calls, 7 mutations identified by TS were not called by uWGS. Six of them were not detected due to filtering criteria based on genomic context, and 1 was not identified by uWGS with a VAF of 13% by TS. To evaluate the utility of uWGS, we performed a detailed review of alterations identified in the 43 samples with unknown or incomplete driver profiles to assess the incremental value of uWGS over standard testing. In 39/43 (91%) samples, uWGS identified at least one additional oncogenic event compared to standard profiling, with 74% (32/43) of cases with incremental clinically relevant findings. Additionally, 8/43 (19%) cases had driver alterations only detectable by uWGS. Alterations with clinical value were detected by uWGS across all mutational types, with deletions and translocations being the most frequent. Small deletions (n=44, median=0.15 mega base pairs) primarily affected tumor suppressors, such as CDKN2A (n=10), ETV6 (n=6) and IKZF1 (n=4). Twenty-one genomic rearrangements were identified solely by uWGS. Nine of 21 rearrangements were identified in cases where RNA fusion-panel was not performed as part of the LEAP study, while the remaining 12 driver rearrangements were from samples that had RNA testing. Moreover, 2 novel translocations, RUNX1::HOXD1 and RUNX1::HOXA13, weredetected in an AML and MPAL, respectively. Conclusions We demonstrate that comprehensive profiling of acute leukemias can be successfully performed using uWGS. Use of this platform allows for return of results with high concordance with standard clinical multi-platform genomic assessments. Importantly, uWGS provides additional clinically relevant information in a substantial percent of acute leukemia samples.