Merck, SARC, Sarcoma Foundation of America, QuadW Foundation, Pittsburgh Cure Sarcoma, and Ewan McGregor.

Purpose: We recently reported a 17.5% objective RECIST 1.1 response rate in a phase II study of pembrolizumab in patients with advanced sarcoma (SARC028). The majority of responses occurred in undifferentiated pleomorphic sarcoma (UPS) and dedifferentiated liposarcoma (DDLPS). We sought to determine whether we can identify immune features that correlate with clinical outcomes from tumor tissues obtained pre-and on-treatment.Patients and Methods: Pretreatment (n ¼ 78) and 8-week ontreatment (n ¼ 68) tumor biopsies were stained for PD-L1 and multiplex immunofluorescence panels. The density of positive cells was quantified to determine associations with anti-PD-1 response.Results: Patients that responded to pembrolizumab were more likely to have higher densities of activated T cells (CD8 þ CD3 þ PD-1 þ ) and increased percentage of tumor-associated macrophages (TAM) expressing PD-L1 pre-treatment compared with non-responders. Pre-treatment tumors from responders also exhibited higher densities of effector memory cytotoxic T cells and regulatory T cells compared with non-responders. In addition, higher density of cytotoxic tumorinfiltrating T cells at baseline correlated with a better progression-free survival (PFS).Conclusions: We show that quantitative assessments of CD8 þ CD3 þ PD-1 þ T cells, percentage of TAMs expressing PD-L1, and other T-cell densities correlate with sarcoma response to pembrolizumab and improved PFS. Our findings support that multiple cell types present at the start of treatment may enhance tumor regression following anti-PD-1 therapy in specific advanced sarcomas. Efforts to confirm the activity of pembrolizumab in an expansion cohort of patients with UPS/DDLPS are underway.

No abstract

Introduction CD73 is a membrane-bound enzyme crucial in adenosine generation. The adenosinergic pathway plays a critical role in immunosuppression and in anti-tumor effects of immune checkpoint inhibitors (ICI). Here, we interrogated CD73 expression in a richly annotated cohort of human lung adenocarcinoma (LUAD) and its association with clinicopathological, immune, and molecular features to better understand the role of this immune marker in LUAD pathobiology. Materials and methods Protein expression of CD73 was evaluated by immunohistochemistry in 106 archived LUADs from patients that underwent surgical treatment without neoadjuvant therapy. Total CD73 (T +) was calculated as the average of luminal (L +) and basolateral (BL +) percentage membrane expression scores for each LUAD and was used to classify tumors into three groups based on the extent of T CD73 expression (high, low, and negative). Results CD73 expression was significantly and progressively increased across normal-appearing lung tissue, adenomatous atypical hyperplasia, adenocarcinoma in situ, minimally invasive adenocarcinoma, and LUAD. In LUAD, BL CD73 expression was associated with an increase in PD-L1 expression in tumor cells and increase of tumor-associated immune cells. Stratification of LUADs based on T CD73 extent also revealed that tumors with high expression of this enzyme overall exhibited significantly elevated immune infiltration and PD-L1 protein expression. Immune profiling demonstrated that T-cell inflammation and adenosine signatures were significantly higher in CD73-expressing lung adenocarcinomas relative to those lacking CD73. Conclusion Our study suggests that higher CD73 expression is associated with an overall augmented host immune response, suggesting potential implications in the immune pathobiology of early stage lung adenocarcinoma. Our findings warrant further studies to explore the role of CD73 in immunotherapeutic response of LUAD.

IntroductionUndifferentiated pleomorphic sarcoma (UPS) can be associated with a relatively dense immune infiltration. Immune checkpoint inhibitors (anti-PD1, anti-PDL1, and anti-CTLA4) are effective in 20% of UPS patients. We characterize the immune microenvironment of UPS and its association with oncologic outcomes.Material and methodsSurgically resected UPS samples were stained by immunohistochemistry (IHC) for the following: tumor-associated immune cells (CD3, CD8, CD163, CD20), immune checkpoints (stimulatory: OX40, ICOS; inhibitory: PD-L1, LAG3, IDO1, PD1), and the adenosine pathway (CD73, CD39). Sections were reviewed for the presence of lymphoid aggregates (LA). Clinical data were retrospectively obtained for all samples. The Wilcoxon rank-sum and Kruskal-Wallis tests were used to compare distributions. Correlations between biomarkers were measured by Spearman correlation. Univariate and multivariate Cox models were used to identify biomarkers associated with overall survival (OS) and disease-free survival (DFS). Unsupervised clustering was performed, and Kaplan-Meier curves and log-rank tests used for comparison of OS and DFS between immune clusters.ResultsSamples analyzed (n=105) included 46 primary tumors, 34 local recurrences, and 25 metastases. LA were found in 23% (n=10/43), 17% (n=4/24), and 30% (n=7/23) of primary, recurrent, and metastatic samples, respectively. In primary UPS, CD73 expression was significantly higher after preoperative radiation therapy (p=0.009). CD39 expression was significantly correlated with PD1 expression (primary: p=0.002, recurrent: p=0.004, metastatic: p=0.001), PD-L1 expression (primary: p=0.009), and CD3+ cell densities (primary: p=0.016, recurrent: p=0.043, metastatic: p=0.028). In recurrent tumors, there was a strong correlation between CD39 and CD73 (p=0.015), and both were also correlated with CD163+ cell densities (CD39 p=0.013; CD73 p<0.001). In multivariate analyses, higher densities of CD3+ and CD8+ cells (Cox Hazard Ratio [HR]=0.33; p=0.010) were independently associated with OS (CD3+, HR=0.19, p<0.001; CD8+, HR= 0.33, p=0.010) and DFS (CD3+, HR=0.34, p=0.018; CD8+, HR=0.34, p= 0.014). Unsupervised clustering of IHC values revealed three immunologically distinct clusters: immune high, intermediate, and low. In primary tumors, these clusters were significantly associated with OS (log-rank p<0.0001) and DFS (p<0.001).ConclusionWe identified three immunologically distinct clusters of UPS Associated with OS and DFS. Our data support further investigations of combination anti-PD-1/PD-L1 and adenosine pathway inhibitors in UPS.