(P 99) M2 MACROPHAGE INFILTRATION IS ASSOCIATED WITH EARLY DISEASE RECURRENCE FOLLOWING CURATIVE INTENT SURGICAL RESECTION IN SOFT-TISSUE SARCOMAS

Objective: The clinical management of soft-tissue sarcomas (STS) is challenging with aggressive surgical resection offering the only chance for cure. Unfortunately, a large proportion of patients develop local recurrence and distant metastasis. Despite advances in therapy for STS, there are no clear clinical guidelines on when or what local or systemic therapy to use in combination with surgical resection or which subsets of patients would benefit from these additional interventions. There is a critical need to improve the diagnosis, prognosis, and treatment of early stage STS using molecular diagnostic and precision medicine approaches. The tumor-immune microenvironment (TIME) represents a vital component of cancer and is an emerging prognostic and predictive biomarker across many malignancies. In STS, the TIME is thought to be cold relative to other cancers; however, recent evidence suggests that subgroups of STS demonstrate a more active TIME. In this study we estimated immune cell infiltration using CIBERSORTx in 69 subjects with early stage STS who received curative intent surgical resection to determine the prognostic value of TIME and identify preliminary biomarkers for targeted therapy.

Methods: Primary, retroperitoneal, formalin-fixed, paraffin-embedded (FFPE) tumor samples were selected from the Mayo Clinic Florida STS Biobank and sequenced using the Tempus xT targeted DNA and whole transcriptome next-generation sequencing platforms. Immune cell deconvolution was calculated using the original CIBERSORT method within the CIBERSORTx deconvolution algorithm. B mode batch correction was applied and the LM22 signature was used as the immune reference. CIBERSORTx was run using both relative and absolute modes with quantile normalization disabled and the deconvolution was estimated for 1,000 permutations. Clinical endpoints included overall survival (OS) and disease-free survival (DFS). OS was defined as time from initial surgical resection to date of death or last known follow-up. DFS was defined as time from initial surgical resection to date of first local recurrence, first metastasis, or last known follow-up. Tumor grade was defined as high (grade 3) or low (grades 1-2). Principle component analysis (PCA) was conducted using the FactoMineRv2.3 package in R. Survival analysis was tested using both log-rank and Cox proportional hazard methods. Unless otherwise stated, p-values ≤0.05 were considered statistically significant.

Results: Of the 69 included STS, the most common histologies included were liposarcoma (n=22), gastrointestinal stromal tumor (n=19), and leiomyosarcoma (n=12). Immune cell infiltration was highly variable across the cohort. Immune infiltration was not associated with tumor purity (p=0.56). The most abundant cell types identified included M2 macrophages, resting T-cells, monocytes, and resting mast cells. Immune infiltration was similar across tissue histologies; however, GIST demonstrated consistently low immune infiltration. Low-grade tumors also demonstrated a lower Immune infiltration compared to high-grade tumors. Using principle component analysis (PCA), we determined that immune infiltrate heterogeneity was primarily driven by the prevalence of absolute immune cells, macrophages (M0 and M2), and monocytes (Figure). Unsurprisingly, M2 macrophages were the highest correlated cell type with absolute immune scores and were the dominant cell type identified across the dataset. While M0 macrophages were not highly prevalent in this cohort, their impact on tumor clustering is likely due to their inverse relationship with M2 macrophages. To determine the correlation between immune cell infiltration and clinical outcomes in this dataset, we took a two-step approach, requiring gene sets to be correlated with both DFS and OS first as categorical variables (median split, p< 0.05), then as continuous variables (z-scores, p< 0.1). Only M2 macrophage infiltration was associated with clinical outcomes with high infiltration associated with poor DFS and OS.

Conclusion: Prior studies have associated STS with cold immune environments; however, with the high infiltration present in a small number of tumors in this dataset, it is possible that there may still be a spectrum of immune environments in STS. We identified a M2 macrophage dominant, pro-tumor TIME in high-risk disease. Taken together, this data may help to identify patients receiving curative intent surgical resections for intra-abdominal STS who are at the highest risk of recurrence, while presenting the potential target for immunotherapy in a subset of high-risk STS.