Unmatched Whole-Genome Sequencing As a Clinical Tool for Hematological Neoplasms with Significant Utility in Cases with Tumor-in-Normal Contamination

Abstract

Introduction For patients with acute leukemias, mapping of clinical biomarkers is critical for disease classification, risk stratification and treatment decisions. At diagnosis, molecular profiling is primarily based on cytogenetic analysis, SNP-arrays and increasingly complemented with targeted assays to include Fluorescent In Situ Hybridization (FISH), and DNA/RNA panel assays. Recently, Duncavage et al. showed that for acute myeloid leukemia, whole genome sequencing (WGS) can replace cytogenetics and detect the hotspot mutations in the ELN2017 guidelines (Duncavage et al., 2021). However, the 2022 revision of disease classification and risk stratification led to the incorporation of additional events characterized by a diverse spectrum of non-hotspot mutations (Khoury et al. 2022, Bataller et al. 2022). To this end, the diagnostic utility of WGS for comprehensive characterization of relevant biomarkers across hematological neoplasms is not established. A further limitation for matched WGS (mWGS) is the requirement of a normal sample at diagnosis, as sources of control tissue can be contaminated with tumor cells. Methods To resolve this unmet clinical need, we developed a computational framework for the analysis of WGS data without the use of a normal germline control (uWGS). We applied this uWGS in 2 independent cohorts of patients (Table 1). In cohort 1 of 49 patients, representative of a broad range of hematological neoplasms, uWGS findings were compared against standard-of-care molecular assays to include cytogenetics, FISH, SNP-arrays, targeted-gene panels and RNA sequencing for findings of clinical relevance (disease classification, risk stratification, therapy informing) as annotated by OncoKB (Chakravarty et al. 2017). To further evaluate whether an uWGS can detect the same spectrum of acquired mutations that a matched WGS workflow (mWGS), we compared findings for 52 B-ALL patients in cohort 2 with mWGS data, and we estimated the tumor-in-normal contamination (TiN) with the algorithm deTiN (Taylor-Weiner et al., 2018). Results In Cohort 1, uWGS detected 100% (24/24) of highly confident findings (cancer cell fraction >20) of clinical relevance described by standard-of-care assays (0.49 average, 0-3 range) (Fig.1). Moreover, uWGS identified 9 mutations in 8 patients that were not detected by standard-of-care. These include 2 disease-defining P2RY8::CRLF2 in B-ALL, 2 canonical t(7;9)(q34;q34.3) NOTCH1-intergenic in T-ALL, a TCF3 D561V point mutation and a IGH::MYC event in Burkitt lymphoma, 2 focal deletions in CDKN2A and IKZF1 in B-ALL, and a focal NF1 deletion in JMML. Importantly, 5/8 of these cases did not have a clinically relevant alteration identified by standard-of-care. In Cohort 2, we evaluated what proportion of relevant events identified by mWGS can be detected by uWGS. Putative biomarkers had been identified by mWGS in 88% of patients (46/52). These included delineations of aberrant karyotypes where chromosome banding failed (5/52), newly described fusion genes (i.e. UBTF::ATXN7L3, EP300::ZNF384, in 27/52) and recurrent gene mutations (i.e. PAX5 P80R, ZEB2 H1038R, in 14/52). As a result, uWGS workflow captured 100% of biomarkers identified in the mWGS (5/5 ploidy, 27/27 fusion, 14/14 coding). Remarkably, evaluation of TiN levels reveals that 4 driver events had been missed in the mWGS workflow, affecting genetic alterations across variant classes. For example, a focal BTG1 loss detected by uWGS was not called by mWGS. Overall, 25% of B-ALL cases had evidence of TiN >2% (14/55, ranges 2-57%), resulting in lower sensitivity of detection of copy number variants (CNVs) in mWGS. When looking into the whole spectrum of CNVs detected only by uWGS, we found lower levels of precision in mWGS that were associated with evidence of TiN (median 0.45 vs 0.64, p-value 0.03). Additionally, the other 3 drivers missed by mWGS were indels in CHEK2, JAK1 and TCF3. Conclusions We have developed and validated an uWGS workflow for the detection of clinically relevant alterations in leukemia across variant classes to include CNVs, translocations and point mutations. We demonstrated that for hematological neoplasms, uWGS rescues events that can be missed by mWGS workflows owing to TiN, which shows how uWGS enables the detection of clinically relevant biomarkers, and the opportunity to discover new clinical findings using a single test and a single biopsy. Figure 1View largeDownload PPTFigure 1View largeDownload PPT Close modal