AUTHORS
Olga Sala-Torra, MD, David Woolston2, Maika Malig, PhD3, Lan Beppu4, Stephen Eacker, Ph.D.5, Cecilia CS Yeung, MD, Professor Pathology6, Min Fang, MD, PhD, FACMG7, Jerald Radich, MD8
AFFILIATIONS
1 Fred Hutchinson Cancer Center, Seattle, Washington, United States
2 Fred Hutchinson Cancer Center, Seattle, Washington, United States
3 CytoTerra, Seattle, Washington, United States
4 Fred Hutchinson Cancer Center, Seattle, Washington, United States
5 CytoTerra, Seattle, Washington, United States
6 Fred Hutchinson Cancer Center, Seattle, Washington, United States
7 Fred Hutchinson Cancer Center, Seattle, Washington, United States
8 Fred Hutchinson Cancer Center, Seattle, Washington, United States
ABSTRACT
Background: Clinical guidelines require cytogenetic evaluation of all acute myeloid leukemia (AML) cases to ascertain patient risk and to tailor their treatment regime. While traditionally cytogenetic evaluation uses some combination of karyotyping, FISH, and array, an increasing number of next-generation methods are available to evaluate the structural variants (SVs) present in patient genomes. In this study, we seek to evaluate two such methods: short-read whole genome sequencing (WGS) and genomic proximity mapping (GPM), benchmarking their performance against current standard of care cytogenetics.
Methods: We collected 48 banked AML patient samples with available cytogenetic data and processed them for WGS and GPM. WGS libraries were prepared using 300 ng DNA with a PCR-free library prep, sequenced to a median depth of 72.5X and analyzed using the DRAGEN Somatic analysis pipeline v4.3.6 in tumor-only mode, and a secondary analysis pipeline with filtering parameters that capture fusion variants affecting genes and regions involved in hematologic malignancies, CNV with length at least 0.5 Mbp and with 20% change in read depth, and cnLOH with size at least 10 Mbp extending to the chromosome end. GPM libraries were generated from 100,000–500,000 cells and sequenced to a median depth of 19X and evaluated on the Curator cloud analysis platform by an analyst blinded to previous findings. Findings of each platform were compared to each other and with clinical cytogenetics findings from the Fred Hutchinson Cancer Center Clinical Cancer Genomics Laboratory.
Results: WGS and GPM demonstrated 100% concordance for specific SV biomarkers identified by the European Leukemia Network (ELN) and World Health Organization guidelines (n = 12). Overall, including ELN-specified and otherwise, WGS corroborated 50/60 SVs detected by GPM and GPM corroborated 50/56 SVs detected by WGS. GPM had a higher concordance rate with clinical findings, identifying 43 cytogenetics-corroborated variants compared to 37 by WGS. Post-hoc manual review found support for two additional concordant variants by GPM and five by WGS, missed due to filter stringency or low coverage. GPM and WGS identified cytogenetically cryptic rearrangements including an inv(3)(p24.3q26.2) involving the known AML driver gene MECOM (EVI1) not observed by cytogenetics. WGS was more sensitive at detecting sub-megabase copy number alterations. Karyotyping was more sensitive to whole chromosome gains or losses and subclonal chromosome abnormalities present at less than 20%. When considering solely SVs, choice of platform did not significantly affect patient risk assessment (χ2 = 0.17, p > 0.99), with seven discordant patient risk assignments compared with standard-of-care cytogenetics in four of the 48 cases. Of these instances of discordance, GPM and WGS were concordant in six of seven assignments.
Conclusion: Both platforms evaluated in this study demonstrate the ability to identify SVs used to guide AML patient care. WGS and GPM represent single, scalable next generation methods for cytogenomic testing which currently require multiple labor-intensive cytogenetic tests. Of these two methods, GPM most closely matched the sensitivity of current standard-of-care while increasing resolution for sub-megabase copy number alterations and cytogenetically cryptic rearrangements.
DISCLOSURES
Olga Sala-Torra, MD: No relevant conflicts of interest to declare.