(P 22) SINGLE CELL SEQUENCING ENABLES LINEAGE-INFORMED TARGETING OF OSTEOSARCOMA

Objective: Osteosarcoma (OS) remains difficult to treat due to its genetic heterogeneity and lack of consistent targetable mutations. Notably, current OS treatment is not subtype-specific, and morphology alone does not reveal targets. Here, we demonstrate that single-cell transcriptomic profiling can resolve conserved differentiation states - or "archetypes" - with prognostic and pharmacologic relevance.

Methods: We used a non-negative matrix factorization-based framework to project bulk RNA-seq data from 16 OS cell lines and six matched primary-relapse patient-derived xenograft (PDX) tumor pairs onto a single-cell atlas of normal mesenchymal differentiation.

To find archetype-selective compounds, we (1) compared transcriptional responses of two OS cell lines to 337 agents via PLATE-seq (a low-depth, multiplexed RNA-seq method), (2) correlated archetype scores with DepMap viabilities, and (3) prioritized clinically relevant drugs. We then tested 32 drugs across six OS cell lines in dose-response assays.

Results: We identified three dominant archetypes: a stem-like mesenchymal state, a committed chondrogenic state, and a transitional osteoblastic state. The stem-like state was enriched in relapsed PDX tumors and associated with poor survival in the TARGET-OS cohort, suggesting a role in therapeutic resistance.

Archetype scores significantly correlated with drug responses (p < 0.05, Spearman rank test). Stem-like lines were differentially sensitive to refametinib (IC50 range: 1.37 - 66.0 µM), a MEK inhibitor, consistent with MAPK pathway dependence. Osteoblastic-like lines were more sensitive to belinostat (IC50 range: 0.30 -1.66 µM), an HDAC inhibitor, reflecting epigenetic plasticity in transitional states.

Conclusion: Our single-cell atlas offers a foundation for lineage-specific therapeutic targeting and disease monitoring. Expanded drug screening and larger-scale in vivo validation studies are needed to inform future precision strategies in OS.