(P 404) PROTEOMIC PROFILING OF THE SOFT TISSUE SARCOMA MICROENVIRONMENT IDENTIFIES A PROGNOSTIC SIGNATURE BASED ON PROTEOGLYCAN EXPRESSION AND CD8+ TUMOUR INFILTRATING LYMPHOCYTES

Objective: Components of the tumour microenvironment (TME), such as the extracellular matrix (ECM) and their cognate adhesion receptors, drive important hallmarks of tumorigenesis, including tumour progression and metastasis. These components remain poorly characterised in soft tissue sarcomas (STS), and there is a limited understanding of their role in STS pathobiology. Here, we undertook a comprehensive proteomic characterisation of the STS TME in a large cohort of patients to identify candidate prognostic biomarkers for risk stratification.

Methods: We employed mass spectrometry to profile the STS TME (including the ECM and their cognate adhesion receptors) in primary tumour specimens from a cohort of 321 STS patients across 11 histological subtypes. This led to the identification of 302 ECM related proteins. Significance Analysis of Microarrays Two-class unpaired tests were utilised to define TME proteins significantly enriched in each histological subtype. To investigate the intra-subtype tumour heterogeneity of the TME within leiomyosarcomas (LMS) (n=80) and dedifferentiated liposarcomas (DDLPS) (n=39), consensus clustering of ECM related proteomic profiles was performed. Univariate Cox regression analysis of single-sample Gene Set Enrichment Analysis (ssGSEA) scores of different TME gene subsets was evaluated to identify candidate prognostic protein signatures. In parallel, tumour infiltrating lymphocyte (TIL) levels (CD3, CD4 and CD8) were assessed by immunohistochemistry in LMS, DDLPS and undifferentiated pleomorphic sarcoma (UPS) (n=148) cases. Multivariable Cox regression analysis was performed to determine whether the ssGSEA prognostic signatures were independent of known prognostic clinicopathological variables.

Results: Unsupervised clustering of the dataset demonstrated that distinct histological subtypes had unique TME proteomic profiles (Figure 1A). Supervised analysis of significantly enriched proteins showed that LMS tumours had upregulated levels of proteins involved in cell adhesion (e.g. ITGA3/5/7 and ITGB1/3) and various glycoproteins (e.g. LAMA4/5, LAMB2, LAMC1, NID1/2, SBSPON) (Figure 1B). LMS and DDLPS showed intra-subtype heterogeneity of TME protein expression with consensus clustering identifying three significant subgroups of patients in each subtype. In the case of LMS, these subgroups were associated with tumour grade, while in DDLPS, the subgroups were prognostic of local recurrence-free survival and overall survival (OS). In a subset of LMS, UPS and DDLSP patients (n=148), integrating a ssGSEA score of 11 proteoglycans together and CD8+ TILs count was prognostic for OS (Figure 1C). Patients with low proteoglycan score/low CD8+ TILs had a significantly worse OS compared to patients with a high proteoglycan score/high CD8+ TILs (univariate analysis: Hazard Ratio (HR)=0.33, 95% Confidence Interval (CI)=0.17-0.65, p-value=0.001). Moreover, after adjusting for tumour grade, size, histological subtype, performance status, sex and age, low proteoglycan score/low CD8+ TILs score remained an independent prognostic factor for OS (multivariable analysis: HR=0.33, 95% CI=0.16-0.70, p-value=0.004).

Conclusion: Our characterisation of the TME in a large cohort of STS patients showed that STS histological subtypes exhibit unique TME protein expression profiles. There is intra-subtype TME heterogeneity in LMS and DDLPS, which is associated with tumour grade and survival outcome measures. Integration of the proteoglycan score with CD8+ TILs counts led to the discovery of an independent prognostic factor for OS in LMS, DDLPS and UPS patients.