Session: 803. Emerging Tools, Techniques, and Artificial Intelligence in Hematology: Poster II
Hematology Disease Topics & Pathways:
Artificial intelligence (AI), Bioinformatics, Emerging technologies, Technology and Procedures, Profiling, Machine learning, Molecular testing, Omics technologies
Chronic lymphocytic leukemia (CLL) is a heterogeneous disease of B lymphocytes with diverse molecular and phenotypic characteristics. Precision medicine approaches have shown promise for personalized treatment selection and disease monitoring in CLL patients. While traditional profiling uses flow cytometry (FC) to identify surface protein patterns, it cannot detect gene signatures for treatment response/resistance, or identify malignant B cell receptor (BCR) clonotypes. To address this gap, we present a novel liquid biopsy-based approach to characterize and monitor CLL disease using next generation sequencing (NGS) of plasma cell-free RNA (cfRNA).
Methods
Whole blood was collected from 34 CLL patients: 20 treatment-naive, 12 on different types of treatments, and 2 post-treatment. Abundance of cfRNA transcripts from cell type-specific signatures was analyzed using single-sample gene set enrichment analysis (ssGSEA). BCR repertoire reconstruction and deconvolution (Zaitsev et al., Cancer Cell, 2022) of major cell populations were performed from peripheral blood mononuclear cell (PBMC) RNA-seq and plasma cfRNA-seq. The fraction of each clonotype was calculated based on sequencing coverage. Proportions of malignant B cells were predicted from cfRNA using our CLL-specific ML model. Flow cytometry staining panels were used to detect malignant B cells and profile blood immune populations from PBMCs. Somatic mutation calling was performed using the BostonGene Tumor Portrait assay on DNA extracted from B cells or PBMCs as the source of malignant cells and from sorted or enriched T cells as the source of normal cells. Tumor-specific mutations were called from cfRNA fractions using Pisces 5.2.
Results
cfRNA-based deconvolution and ssGSEA revealed cell composition dynamics relevant for treatment monitoring. Significant changes were detected starting from the first weeks of therapy: decreased B cell levels (P=0.00003, Mann-Whitney U test) on- and post-treatment, increased NK cells (P=0.00006), and no significant changes in T cells or monocytes. Moreover, some tissue cell populations such as macrophages (P=0.00002), endothelium (P=0.00001), fibroblasts (P=0.03), and hepatocytes (P=0.002) showed significantly increased levels on- and post-treatment.
cfRNA-inferred dominant BCR clones for cases with high tumor cell fraction detected by FC were considered malignant. Among 20 patients with both cfRNA-seq and PBMC RNA-seq available, 18 patients had dominant clones matched by heavy chain CDR3 sequences and 17 patients had dominant clones matched by light chain CDR3 sequences. Comparison of the coverage of the BCR region for malignant clonotypes relative to protein-coding regions showed 2 times (P=0.008, Wilcoxon test) higher levels in cfRNA compared to PBMC RNA-seq indicating potential advantages for monitoring treatment, relapse, and minimal residual disease (MRD) using cfRNA.
To directly predict the malignant B cell fraction from cfRNA, we developed a CLL-specific ML model. The model predictions correlated well with cfRNA-derived malignant B cell fraction, calculated by multiplying the total deconvolved B cell fraction by the fraction of the dominant BCR clone (Pearson correlation=0.83, P=0.01). Model outputs divided by the fraction of total B cell populations also correlated with the malignant BCR clonotype fractions (Pearson correlation=0.71, P=0.05), indicating the model’s ability to distinguish between malignant and healthy B cells.
Finally, we detected clinically significant tumor-derived mutations from cfRNA. In total, 24 coding mutations were identified and confirmed by WES, of which 9 were clinically significant. These somatic variants included LoF mutations in TP53, ATM and BIRC3, mutations in SF3B1, truncating mutations in the PEST domain of NOTCH1, and several others.
Conclusions
The presented liquid biopsy-based approach demonstrates the feasibility of identifying malignant cell fraction, BCR repertoires, clinically significant mutations, and immune and tissue specific processes from cfRNA. This study also suggests that the cfRNA platform may support longitudinal monitoring of CLL progression and relapse, as well the development of MRD tests with future validation.
Disclosures: Radko: BostonGene: Current Employment. Kwak: BostonGene: Ended employment in the past 24 months. Shubin: BostonGene: Current Employment, Current holder of stock options in a privately-held company, Patents & Royalties: patents. Krauz: BostonGene: Current Employment, Current equity holder in private company, Current equity holder in publicly-traded company, Current holder of stock options in a privately-held company, Patents & Royalties: patents. Zaitsev: BostonGene: Current Employment, Current equity holder in private company, Current equity holder in publicly-traded company, Current holder of stock options in a privately-held company, Patents & Royalties: patents. Tarabarova: BostonGene: Current Employment. Savchenko: BostonGene: Current Employment, Current equity holder in private company, Current holder of stock options in a privately-held company, Patents & Royalties: patents. Meerson: BostonGene: Current Employment. Bolshakov: BostonGene: Current Employment. Lupova: BostonGene: Current Employment. Lugovykh: BostonGene: Current Employment. Nikitin: BostonGene: Current Employment, Current equity holder in publicly-traded company, Divested equity in a private or publicly-traded company in the past 24 months. Pashkovskaia: BostonGene: Current Employment. Merriam: BostonGene: Current equity holder in private company, Current holder of stock options in a privately-held company, Ended employment in the past 24 months. Abdou: BostonGene: Current Employment, Current equity holder in private company, Current holder of stock options in a privately-held company. Zheleznyak: BostonGene: Current Employment. Markova: BostonGene: Current Employment. Kersilova: BostonGene: Current Employment. Yudina: BostonGene: Current Employment, Current equity holder in private company, Current holder of stock options in a privately-held company. Shpak: BostonGene: Current Employment, Current equity holder in private company, Current holder of stock options in a privately-held company. Kamysheva: BostonGene: Current Employment, Current equity holder in private company, Current holder of stock options in a privately-held company, Patents & Royalties: patents. Terenteva: BostonGene: Current Employment. Fowler: BostonGene: Current Employment, Current equity holder in private company, Current holder of stock options in a privately-held company; CelGene: Consultancy, Research Funding; Roche/Genentech: Consultancy, Research Funding; TG Therapeutics: Consultancy, Research Funding; Bayer: Consultancy; Novartis: Consultancy, Research Funding; Verastem: Consultancy; Gilead: Research Funding; Abbvie: Research Funding; BeiGene: Research Funding. Louissaint: Massachusetts General Hospital: Research Funding. Goldberg: BostonGene: Current Employment, Current equity holder in private company, Current holder of stock options in a privately-held company, Patents & Royalties: patents.