Type: Oral
Session: 614. Acute Lymphoblastic Leukemias: Biomarkers, Molecular Markers, and Minimal Residual Disease in Diagnosis and Prognosis: Understanding and Exploiting Molecular Therapeutic Targets in ALL
Hematology Disease Topics & Pathways:
Research, Lymphoid Leukemias, ALL, Acute Myeloid Malignancies, AML, Translational Research, Bioinformatics, Diseases, Lymphoid Malignancies, Computational biology, Technology and Procedures, Measurable Residual Disease , Machine learning, Omics technologies
In several subtypes of acute lymphoblastic leukemia (ALL), lymphoblasts often switch to a myeloid (mostly monocytic) lineage during the early phase of treatment. This unusual but well-documented phenomenon can cause diagnostic problems, particularly in the assessment of response to treatment. The cause of the lineage switch and its dynamics are not yet fully understood. Modern multiomic methods provide a unique opportunity to elucidate the progression of B-to-myeloid transdifferentiation and to clarify how different subtypes of ALL with a propensity for lineage instability differ. Previous studies have focused only on blasts at diagnosis (Dx); in our study, we also incorporated blasts from peripheral blood (PB) during or after the initial corticosteroid prephase of treatment.
Methods:
We used 10x Genomics’ Single Cell 5' v2 platform to simultaneously evaluate gene expression and cell surface protein expression at the single-cell level from 3 pediatric patients with different subtypes of ALL prone to phenotypic instability (DUX4 rearrangement, ZNF384 rearrangement, and PAX5 P80R mutation). Blasts from bone marrow (BM) or PB from Dx were analyzed together with PB from day 8 (n=1) or day 5 (n=1) of a BFM-type treatment protocol. The samples were stained with a panel of TotalSeq-C antibody-oligonucleotide conjugates and processed using the Single Cell 5' workflow. Sequencing was conducted on an Illumina NextSeq2000 platform. A mass cytometry panel combining myeloid and lymphoid markers was used to measure protein expression in paired samples (Dx and on-treatment sample) from patients with a B-to-myeloid switch with DUX4r (n=3), PAX5 P80R (n=1), and ZNF384r (n=1) subtypes.
Both RNA and protein counts were derived with a pipeline based on The Single-cell Pediatric Cancer Atlas project (Hawkins, bioRxiv, doi: 10.1101/2024.04.19.590243) and further data processing, including background removal and doublet detection, was carried out in accordance with current recommendations (Heumos, Nat Rev Genet 2023). Our trajectory inference framework, tviblindi (Stuchly, eLife 2024), was used to infer dynamic processes from the static snapshots provided by single-cell measurements.
Results:
In all subtypes, we were able to detect an intermediate population connecting B- and switched blasts, favoring a model in which the switch occurs via transdifferentiation. In the PAX5 P80R patient, we observed an abrupt change to a myeloid phenotype on day 5 of treatment. However, we identified the intermediate population at Dx on a transcriptomic level. In the ZNF384r patient, lineage instability was already evident at Dx, with a portion of the blasts showing differentiation towards the myeloid lineage while harboring a clonal immunoglobulin heavy chain (IGH) rearrangement on a DNA level as determined from sorted myeloid blasts. The expression of this IGH rearrangement decreased in myeloid blasts together with IGHM. In the DUX4r patient, B-to-myeloid transdifferentiation was detected on day 8 of the corticoid pre-phase, accompanied by a smaller B lymphoid blast population showing increased protein expression of CD45 and an altered transcriptome distinguishing this cluster from B lymphoid blasts at Dx.
The immunophenotype of the intermediate populations was typical of B-blasts, but a shared set of myeloid genes was upregulated in their transcriptomes (e.g., CSF2RA, CEBPD, CD86, CD68, CYBB, S100A4, TYROBP, SRGN). Trajectory inference analysis was consistent with a transdifferentiation path through these intermediate populations in all samples, with an expected lag between transcriptomic and immunophenotypic alterations. Notably, CD371 was upregulated at the onset of the switch, along with other myeloid markers.
Summary:
Although the dynamics of the switch varied between subtypes, we observed an intermediate cellular population of varying size in patients with DUX4r, ZNF384r, and PAX5 P80R ALL, which was immunophenotypically consistent with B-blasts while sharing a common myeloid-primed transcriptomic signature. Furthermore, pseudotemporal ordering with tviblindi placed these intermediate populations before the emergence of monocytic cells in all samples, supporting our transdifferentiation hypothesis.
Supported by NU23-05-00353, NW24-07-0026, NU23J-03-00026, UNCE/24/MED/003, and LX22NPO5102.
Disclosures: No relevant conflicts of interest to declare.