(P 20) IDENTIFICATION OF A PROGNOSTIC GENE SIGNATURE FOR OSTEOSARCOMA RELAPSE BASED ON RESIDUAL TUMOR DISEASE AND ITS MICROENVIRONMENT

Objective: The prognosis of osteosarcoma (OS) patients is largely dependent on their response to chemotherapy, yet the evaluation method for this response is not optimal. We hypothesized that the residual tumor after chemotherapy could help identify a prognostic expression signature for relapse in OS.

Methods: To address this hypothesis, we mapped the heterogeneity of treatment responses in different regions of resected samples using single-nucleus RNA sequencing (snRNA-seq). We performed snRNA-seq on 9 samples from 7 patients (3 poor responders, 4 good responders). We identified microenvironment populations and residual tumor cells. For each subset of the population (TILs, osteoblast-like cells, myeloid cells, pericytes, CAFs, and endothelial cells), we selected the top 10 differentially expressed genes from each cluster. We evaluated prognostic genes for progression-free survival (PFS) using univariate Cox regression models from an original RNA-seq "bulk" cohort (n=75), annotated by an expert bone pathologist. We then computed centroids for each sample using a leave-one-out cross-validation approach to classify our cohort into two groups. The prognostic value of our signature was confirmed using Kaplan-Meier survival analysis for PFS. We validated our signature by performing survival analysis on the TARGET cohort (n=80) by transferring the centroid values to determine our groups.

Results: We identified a total of 12 genes with prognostic value for PFS using univariate Cox regression models. Our signature demonstrated a prognostic value for PFS (p < 0.00055) in our original cohort (n=75). This signature was validated in the TARGET cohort (n=80) and was prognostic for event-free survival (p = 0.025) and overall survival (p = 0.019).

Conclusion: Based on the analysis of residual disease, we generated a prognostic signature derived from the tumor microenvironment and residual tumor cells. This signature of treatment resistance, prognostic from the biopsy stage, could improve the therapeutic management of osteosarcoma.