Paper Details

ABSTRACT

Introduction: Cytogenetic abnormalities are central to AML diagnosis and ELN risk classification, yet standard testing requires multiple complementary assays and may fail to detect clinically relevant cryptic structural variants and copy-neutral loss of heterozygosity. Standard cytogenetic testing includes metaphase karyotyping, fluorescence in situhybridization (FISH), and occasionally chromosome genomic array testing (CGAT). This combination of testing is used to capture the breadth of actionable and prognostic chromosome aberrations observed in AML. However, current cytogenetic practices can fail to identify at risk patients due to inadequate resolution or incomplete testing (i.e. failed karyotype or no CGAT) and turnaround time is variable.

Methods: We evaluated whether genomic proximity mapping (GPM), a specialized low-pass whole genome Hi-C sequencing approach, could comprehensively identify clinically relevant structural abnormalities in AML. We performed a multi-center analysis of retrospectively banked diagnostic AML patient samples (n = 206), including patients from the SWOG trial S0106, and evaluated variants detected with GPM compared to clinical cytogenetics findings. Discordant findings were reassessed using appropriate orthogonal methods to establish performance statistics. Using the ELN 2022 risk stratification algorithm and available single nucleotide variant (SNV) panel sequencing data, we performed risk stratification of the patient population using cytogenetic or GPM variant data.

Results: Findings between cytogenetics and GPM were highly correlated with an overall positive percent agreement (PPA) of 0.90 and a negative percent agreement (NPA) of 0.97.  GPM detected numerous clinically relevant lesions missed by conventional cytogenetics, including four cryptic KMT2A rearrangements, eight NUP98 rearrangements, and seven cases of 13q copy-neutral LOH. Conversely, GPM demonstrated a 0.73 PPA for whole chromosome aneuploidy, dominated by discordant +8 cases, an abnormality that commonly yields growth advantage during myeloid karyotyping culture. GPM identified every instance of -7, an adverse risk biomarker.

Using outcomes for these patients and the ELN2022 guidelines, we evaluated risk stratification utilizing GPM or cytogenetics to identify relevant cytogenetic-scale abnormalities. The results were nearly identical  between cytogenetics (favorable vs intermediate: HR=0.60, 95% CI 0.36−1.01, p=0.05, adverse vs intermediate: HR=1.53, 95% CI 0.99−2.38, p=0.05) and GPM (favorable vs intermediate: HR=0.58, 95% CI 0.35−0.98, p=0.04, adverse vs intermediate: HR=1.53, 95% CI 0.98−2.38, p=0.06). Incorporating 13q cnLOH into the ELN2022 framework creates an improvement on the segregation of adverse risk patients (cyto: adverse vs intermediate: HR=1.35, 95% CI 0.86−2.10, p=0.2 versus GPM: adverse vs intermediate: HR=1.79, 95% CI 1.14−2.80, p=0.01).

GPM identified cnLOH of 13q to be associated with poor outcomes (HR = 2.06, p =0.04). Cytogenetic testing failed to observe this trend, likely because of incomplete application of CGAT testing to the cohort (n = 38 cases with CGAT). Of note, all observed instances of 13q cnLOH were accompanied by FLT3-ITD mutations, indicating the homozygous status of FLT3-ITD in these patients.

Conclusions: These data suggest that GPM and cytogenetic assay results are highly correlated and GPM can accurately classify patient risk using ELN2022 guidelines. GPM provides patient risk stratification more efficiently by providing a single comprehensive cytogenomic test that is a sensitive detector of cytogenetically cryptic variants. Notably, every case of 13q cnLOH occurred in FLT3-ITD AML, consistent with duplication of the mutant allele. Incorporation of this lesion significantly improved adverse-risk stratification and suggests that GPM identifies clinically relevant genomic events not routinely captured by current diagnostic workflows.