Expression and prognostic significance of PDGF ligands and receptors across soft tissue sarcomas.

BACKGROUND: While the anti-PDGFRA antibody olaratumab failed to confirm an impact on survival in unselected advanced soft tissue sarcoma (STS) patients, the level of expression and the prognosis of platelet-derived growth factor (PDGF) receptors and ligands in STS remain unclear. PATIENTS AND METHODS: We analyzed PDGF ligands and receptors' expression levels in a series of 255 patients with different histologies of STS [gastrointestinal stromal tumor (GIST), myxoid liposarcoma (MLPS), sarcoma with complex genomics, synovial sarcoma (SyS)] with Agilent single-color micro-arrays. We explored expression levels as prognostic values in univariate and multivariate analysis using R software (version 3.4.2). RESULTS: Complex patterns of correlation of expression between ligands and receptors were observed for each histotype. PDGFA levels were highest in SyS and lowest in MLPS (P < 4 × 10-9), PDGFB and C levels were lower in GIST (P < 2 × 10-15 and P < 3 × 10-9) while PDGFD expression was similar across histological subtypes. PDGF receptor (PDGFR) A expression was lowest in MLPS (P < 0.002), whereas PDGFRB and L expressions were lowest in GIST and SyS (P < 0.0004). Interestingly, high PDGFA expression levels were associated with higher risk of metastasis (P = 0.006), whereas PDGFD levels above average were associated with a reduced risk of metastasis (P = 0.01) in univariate and multivariate analysis. CONCLUSIONS: The expression of PDGF ligands and receptors varies across sarcoma histological subtypes. PDGFA and D expression levels independently and inversely correlate with the risk of metastatic relapse.

Specific immune landscapes and immune checkpoint expressions in histotypes and molecular subtypes of sarcoma.

Soft tissue sarcomas are a group of rare and aggressive connective tissue neoplasms for which curative therapeutic opportunities are limited in advanced phase. Clinical trials assessing immunotherapy in these tumors have so far reported limited efficacy. The objective of this study is to provide a description of the immunologic landscape of sarcomas to guide the next clinical trials of immunotherapy in these diseases. The gene expression profile of 93 immune checkpoint (ICP) and membrane markers (MM) of immune cells was analyzed in a series of 253 soft tissue sarcoma (synovial sarcoma, myxoid liposarcoma, sarcoma with complex genomic and GIST) using Agilent Whole Human Genome Microarrays. The unsupervised hierarchical clustering of gene expression level was found able to properly group patients according to the histological subgroup of sarcoma, indicating that each sarcoma subgroup is associated with a specific immune signature defined by its gene expression pattern. Using the prognostic impact of CIBERSORT signature on metastatic-free survival in each subgroup, specific target could be proposed for each of the four groups: Treg through ICOS and GITR in GIST, M0 macrophages in all four sarcoma subtypes, OX40 in SS, CD40 in GIST and SS. The immune landscape of sarcoma was found to be as heterogeneous as the histotypes and molecular subtypes, but strongly correlated to the histotype. Histotype adapted immunotherapeutic approaches in each sarcoma subtypes must be considered in view of these results, consistently with the already reported specific response of histotypes of ICPs.

Validation of the Complexity INdex in SARComas prognostic signature on formalin-fixed, paraffin-embedded, soft-tissue sarcomas.

Background: Prediction of metastatic outcome in sarcomas is challenging for clinical management since they are aggressive and carry a high metastatic risk. A 67-gene expression signature, the Complexity INdex in SARComas (CINSARC), has been identified as a better prognostic factor than the reference pathological grade. Since it cannot be applied easily in standard laboratory practice, we assessed its prognostic value using nanoString on formalin-fixed, paraffin-embedded (FFPE) blocks to evaluate its potential in clinical routine practice and guided therapeutic management. Methods: A code set consisting of 67 probes derived from the 67 genes of the CINSARC signature was built and named NanoCind®. To compare the performance of RNA-seq and nanoString (NanoCind®), we used expressions of various sarcomas (n = 124, frozen samples) using both techniques and compared predictive values based on CINSARC risk groups and clinical annotations. We also used nanoString on FFPE blocks (n = 67) and matching frozen and FFPE samples (n = 45) to compare their level of agreement. Metastasis-free survival and agreement values in classification groups were evaluated. Results: CINSARC strongly predicted metastatic outcome using nanoString on frozen samples (HR = 2.9, 95% CI: 1.23-6.82) with similar risk-group classifications (86%). While more than 50% of FFPE blocks were not analyzable by RNA-seq owing to poor RNA quality, all samples were analyzable with nanoString. When similar (risk-group) classifications were measured with frozen tumors (RNA-seq) compared with FFPE blocks (84% agreement), the CINSARC signature was still a predictive factor of metastatic outcome with nanoString on FFPE samples (HR = 4.43, 95% CI: 1.25-15.72). Conclusion: CINSARC is a material-independent prognostic signature for metastatic outcome in sarcomas and outperforms histological grade. Unlike RNA-seq, nanoString is not influenced by the poor quality of RNA extracted from FFPE blocks. The CINSARC signature can potentially be used in combination with nanoString (NanoCind®) in routine clinical practice on FFPE blocks to predict metastatic outcome.