(P 426) GD2 HETEROGENEIC EXPRESSION IN OSTEOSARCOMAS AND EWING SARCOMAS UNDERLYING A WORST PROGNOSIS

Objective: Ewing’s sarcomas and osteosarcomas are currently the most common primary bone cancers in adolescents and young adults. Despite various therapeutic approaches, some of these patients develop resistance mechanisms and relapse, leading to tumor progression and the development of metastases. A better understanding of these bone diseases from both intrinsic cellular/molecular and microenvironment extrinsic perspective would enable resistance mechanisms to be targeted more effectively, thereby improving patients' prognosis. These two diseases differ in terms of their cellular origin and genetics, but also in terms of their immune microenvironment. Ewing's sarcoma, composed of round tumor cells from primitive neuro-ectodermal cells originating from neural crests and mesenchymal stem cells, has a fusion transcript and a "warm" immune infiltrate comprising both CD8+ T lymphocytes and macrophages. Osteosarcoma, on the other hand, arises from malignant osteoblastic cells also originating from mesenchymal stem cells, producing an abnormal osteoid matrix, and presents an immune infiltrate composed solely of macrophages. Although these bone sarcomas seem to differ, commonly they are initiating and driven by a mesenchymal stromal cell pattern involving specific TAMs and transversal expression of GD2, a disialoganglioside. This antigen, present to a limited extent in physiological conditions, is involved in cell and tumor promotion, migration and is amenable to targeted therapies. The aim of this work is to look spatially at the distribution of GD2 expression in samples from patients suffering from these two bone sarcomas. We integrate the heterogeneity features of its expression in survival statistical analyses to highlight its impact and the predictive role of this tumoral membrane biomarker.

Methods: A large collection of 28 osteosarcomas and 32 Ewing sarcomas was dedicated to this project after obtaining from patients/legal representatives an informed and signed consent. The immunohistochemical slides were prepared using a microtome (Microm HM 355S, Thermo ScientificTM) to cut the paraffin-embedded samples. We used an immunofluorescent approach similar to immunohistochemistry to estimate GD2 cell membrane expression associated to DAPI nuclear staining. Images were digitalized using a Hamamatsu NanoZoomerS60Digital slide scanner and analyzed with NDP.view2 Viewing software. An experienced pathologist was annotating the region of interest (ROI) inside the samples. The final quantification of positivity was performed using a computer guided alignment followed by QuPath software 0.4.3 analysis.

Results: Our results confirmed the frequent overexpression of GD2 in the Ewing's sarcomas or osteosarcomas’ samples compared with negative controls (healthy lung and colon). Despite the highly heterogeneous localization of GD2 both intratumoral and intertumoral, we were able to quantify globally, evaluate and compare the spatial distribution of GD2 in these samples. The global high GD2 expression was significantly linked to a worst outcome in both bone cancers (ROC analyses determined the GD2 positivity a cut-off of 12%). The automated NDP.view2 and Qupath technologies allow to map spatially GD2 expression in all samples and highlight a significant correlation between a massive heterogeneity driving the protumoral macrophage infiltration in samples.

Conclusion: Those data are providing rationale to consider GD2 as a new marker of prognosis in bone cancers and a way to envision easily Ewing sarcoma and osteosarcoma intra-tumoral heterogeneity. Furthermore, this GD2 overexpression will be rapidly amenable to a substantive targeting.